see rejected  2820070

@@ -9,0 +10 @@
+<s_inf>also written as 盤谷</s_inf>

Koj, Daijr (merge), GG5, etc.

Merging 1617460.

@@ -8,0 +9,3 @@
+</k_ele>
+<k_ele>
+<keb>廻覧</keb>
@@ -20 +23,2 @@
-<gloss>circulation</gloss>
+<gloss>circulation (esp. documents)</gloss>
+<gloss>sending round</gloss>

I can see balance working in some contexts, but no need for it as a gloss.

I don't see how "balance" fits. Not in the JEs.

@@ -22,2 +22 @@
-<gloss>change (i.e. money)</gloss>
-<gloss>balance</gloss>
+<gloss>change (for a purchase)</gloss>

uncontroversial and sitting

koj, daij

@@ -11,0 +11,3 @@
+<k_ele>
+<keb>御釣</keb>
+</k_ele>
@@ -17,2 +20,3 @@
-<xref type="see" seq="1434040">釣り</xref>
-<xref type="see" seq="1434040">釣り</xref>
+<xref type="see" seq="1434040">釣り・2</xref>
+<xref type="see" seq="1434040">釣り・2</xref>
+<misc>&pol;</misc>

Daijs

It's a class of POSs; not a POS itself.

@@ -14 +14 @@
-<gloss g_type="expl">part of speech in Japanese grammar consisting of words that are not particles</gloss>
+<gloss g_type="expl">free-standing Japanese language elements (not auxiliaries or particles)</gloss>

daijr: "国文法での単語の二大別の一。ある観念を表し，文中で単独でも文節を構成することのできる語をいう。名詞・代名詞・動詞・形容詞・形容動詞・連体詞・副詞・接続詞・感動詞の類"
https://ja.wikipedia.org/wiki/品詞#日本語の品詞

Only used in the context of Japanese grammar. Gloss might need a bit of work.
I don't think "substantive" is right. 自立語 includes more than nouns.

@@ -12,2 +12,3 @@
-<gloss>independent (freestanding) word</gloss>
-<gloss>substantive</gloss>
+<field>&ling;</field>
+<gloss>independent word</gloss>
+<gloss g_type="expl">part of speech in Japanese grammar consisting of words that are not particles</gloss>

daijs, gg5

Japanese grammar term.

@@ -13 +13,2 @@
-<gloss>ancillary (attached, adjunct) word or words</gloss>
+<gloss>dependent word</gloss>
+<gloss g_type="expl">auxiliaries and particles in Japanese</gloss>

https://en.m.wikipedia.org/wiki/Telicity

I don't think those additions are very useful.

As for me, I don't mind having those notes. But the question of usefulness for the average user remains.
That said, I believe there's no easier way/term to explain this with, the english wikipedia article on telicity is certainly enough to understand the implications of this note.
What do others think?

登っている (atelic) -> to be climbing
上がっている (telic) -> to have raised

Probably worth adding to the note?

@@ -30 +30 @@
-<s_inf>after the -te form of a verb; indicates continuing action</s_inf>
+<s_inf>after the -te form of an atelic verb; indicates continuing action</s_inf>
@@ -37 +37 @@
-<s_inf>after the -te form of a verb; indicates continuing state</s_inf>
+<s_inf>after the -te form of a telic verb; indicates continuing state</s_inf>

I agree that having "to have ...-ed" first is far more typical for this sense. 
That "construction phrasing" is to my knowledge the standard used in grammar books, and is also taught this way in school.
On its own "to be ...-ed" were more likely to be associated with the passive, that is true, but I think that together with the note the implication remains clear. As nonさん mentioned, "(to) be" also used to be a perfect helper.
Hence one could translate 座っている as "to be seated", which does sound rather formal, but I have nonetheless certainly heard before.

I also agree that it is outright impossible to account for the immense amount if irregularities present in English verb inflection. Hence keeping the sense short but succinct being a good option.

Lastly I believe that having this sense is an improvement one way or another. As someone who has little time to devote to actually studying Japanese, the fact that the aspect indicated by (て)いる varies depending on the telicity of the given verb has long eluded me. So having this in JMdict is overdue in any case.
I believe that a Japanese learner with a sufficient knowledge of the English language will understand what sense four represents.

Apologies for cluttering the feed, but I must correct myself: in line 20 (3rd line from the bottom up) I said 'past participle', I meant to say 'passive voice'.
Just so this is not complete clutter, I adopted the suggestion of shifting the order of 'have -ed' and 'be -ed'.

@@ -37,0 +38 @@
+<gloss>to have ...-ed</gloss>
@@ -39 +39,0 @@
-<gloss>to have ...-ed</gloss>

I would not get too caught up in cases like すわる versus 'to sit'.
Were language an exact science, we would have "I have sat" as a direct equivalent which, as pointed out, is not the only option and 'I am sitting' is just as acceptable and arguably more so.
However, this ambiguity is due to the verb 'to sit', and is not an English generalisation: once again, you cannot take 死んでいる to be 'dying' or 至っている to be 'arriving'.

The matter is that the Japanese すわる acts like a telic verb by default, while 'to sit' does not.
'To sit' patterns like an atelic, dynamic verb like 'to run' in most cases, e.g. you can say "I am running" and "I am sitting" as well as "I sat for an hour" or "I ran for an hour" but not "I sat/ran in an hour" while a verb like 'to die' (which like the Japanese 死ぬ, is telic) assumes a rather different meaning when made into "I am dying" or "I died for an hour", and can be used with an in-adverbial very easily.

With a verb like 'to die' -ing shifts the meaning of the verb from "undergoing death" to "approaching death", while the for-adverbial refers to the duration of the post-state following the verb (in cases where it even applies), rather than the duration of the event itself like with 'sat' or 'ran'.

But then again, "most cases".
"I ran in an hour" for example, can be used so long as either the rest of the phrase or the context provides a delimited something to be run, such as a kilometre, which would then change the verb from being atelic to telic.
'To sit' too has its peculiarities in that "I am sat" means the same as "I am sitting" in most contexts - and note here that 'am' is also an old perfect auxiliary (famously, there is the phrase "I am become death...") though it could be argued that most people do not think of it as such, and that it has its own particular nuance when compared to 'have', e.g. "I am gone" versus "I have gone". 
And then, even 死んでいる which unarguably should be a perfect by default, can be reinterpreted as a continuous, iterated event in phrases like 「そして、おびただしい数の生命が、毎日、死んでいる」。
...Also, could it not be held that the previous "Sat for an hour" instead of the referring to the duration of an event, is instead referring to the duration of the post-state to act of sitting, just like "Died for an hour" refers to the post state to the act of dying? Maybe "I am sitting" is the odd one here...

All this is to reiterate what I opened with: there is too much aspectual doohickey to turn "mismatches" like 座っている　and 'sitting' into a central issue, aspect is not inflexible and a verb can often belong to several categories, or have their categories completely altered by adverbials and context.

About 'be -ed' as a gloss: I included it because of cases wherein, putting the debate on nuances aside, the auxiliary tends to and/or can be 'to be' instead of 'to have', for example "He has gone to tend to family matters" and "He is gone to tend to family matters", etc.
That it is the same configuration used for the past participle, is an unfortunate coincidence (or perhaps it is not, but I digress).

Lastly, regarding "have/be -ed" excluding other past participles: well, that is the nature of irregular verbs. For multiple reasons, we cannot list every single past participle, and what that leaves us with is to indicate the regular form. But I do not think that to be a problem as 'have -ed' really stands for a formula of 'have + past-participle', and applying it to a verb like 'to fall' has the output of 'have fallen'.

座っている, being formed from a 瞬間動詞, falls under the 4th sense of 〜ている, which indicates a resulting state from a completed action. However, depending on context could be translated as "to be sitting" (state, not continuing action) and not "to have sat." English translations are pretty ambiguous when it comes between these two senses.

Additionally, in the 4th sense "to be ...-ed" can easily be mistaken for passive voice, which is misleading. I think it should either come second or be removed. It also assumes regular "-ed" verb forms in English, ignoring many natural equivalents like "to have sat", which but doesn't follow the "-ed" pattern.

In truth, dictionaries - including the Daijirin I used - often even distinguish between subtypes of perfects, but I do not think much is gained from the distinction as they are variations upon the same theme.
Viewing the relation between habituals and progressives the same way, I would not get into habituality as just leaving it with the "be . . .-ing" gloss is probably enough, since a reader faced with 「毎日パンを食べている」would render "I am eating bread every day" and that would be readily understood as habitual. 

In contrast, trying to translate 「死んでいる」 as "dying" is quite distinct from the "died/dead" that it actually means.
And lastly, it is worthy noting that an habitual reading must be forced by either an adverbial or context. Without either「パンを食べている」defaults to being a single-instance progressive, unlike「死んでいる」 which forces a perfect reading because of the meaning of the verb itself.

This is further substantiated by other references as well, like shinsen, smk or meikyo.

Smk even has three senses, having included a sense for habituality as well:
㊀動作・作用が、継続・進行して、その時点に至ることを表わす。
「今手紙を書いて━ところだ」
「泣いて━のか笑って━のか分からない」
㊁動作・作用が反復的に行なわれることを表わす。(Meikyo has this as well)
「毎朝パンを食べて━」
「欠かさず日記を付けて━」
㊂動作・作用が完了したままの状態であることを表わす。
「ドアが開いて━」
「あの人とはどこかで会って━」
「仕事でとても疲れて━」

.

I would be in favor of having separate senses for progressive and perfect usage. It's more tidy and possibly clearer to users while also in accordance with references.

As for the habituality sense in some refs, I'm not sure if or how it should be included. Maybe someone else has an opinion on this?
AFAIK it isn't obligatory to use ている when talking about habits or routines, but I could be wrong

[v1,aux-v] ⇒ [aux-v,v1]

@@ -26,0 +27 @@
+<pos>&aux-v;</pos>
@@ -27,0 +29,5 @@
+<misc>&uk;</misc>
+<s_inf>after the -te form of a verb; indicates continuing action</s_inf>
+<gloss>to be ...-ing</gloss>
+</sense>
+<sense>
@@ -28,0 +35 @@
+<pos>&v1;</pos>
@@ -30,2 +37 @@
-<s_inf>after the -te form of a verb; indicates continuing action or state</s_inf>
-<gloss>to be ...-ing</gloss>
+<s_inf>after the -te form of a verb; indicates continuing state</s_inf>

Daijirin: 「㋐主体の動きを表す動詞に付いて、その動きが継続・進行中であることを表す。」is distinct from「㋑主体の変化を表す動詞に付いて、その結果が持続していることを表す。」

"Be -ing" does not fit as a translation for verbs with a defined terminal endpoint and ensuing post-state such as 死ぬ, 消える, 終わる, etc. wherein the -ている auxiliary does not indicate a progressive and thus cannot be translated as -ing.
E.g. 死んでいる means either "is dead" or "has died", and not "is dying"; 終わっている means "is finished" or "has ended" and not "is ending", etc.
This is a peculiarity of the ている auxiliary in that it doubles down as both the progressive and perfect marker, varying according to the verb, whereas English has a different strategy for each independently of the verb.

I have also removed "have been -ing" as that is perfect progressive, which does not correspond to ている that is either perfect or progressive, but not both.
(That, and the English perfect progressive is not always in practice just what you would in theory get from a simple sum of the perfect plus the progressive, but I digress)

Alternatively, a separate sense could be made for the perfect:
[3][v1,aux-v][uk]
  [note="after the -te form of a verb; indicates continuing action"]
  to be ...-ing
[4][v1,aux-v][uk]
  [note="after the -te form of a verb; indicates continuing state"]
  to be ...-ed; to have ...-ed

@@ -32 +32,2 @@
-<gloss>to have been ...-ing</gloss>
+<gloss>to be ...-ed</gloss>
+<gloss>to have ...-ed</gloss>

I'd leave the [rk].

@@ -7 +6,0 @@
-<ke_pri>ichi1</ke_pri>

BCCWJ:
 いる-非自立  iru	968,064   
 いる         iru	95,901   
 居る         iru	5,247   

5,247/95,901 = 0.055 = 5.5%


Massif for が居る
https://massif.la/ja/search?q=が居る
Out of the first 100 results 3 sentences use 居

I'd still argue that it is rather rare. 
I can see, however, how it could come into conflict with our policy. 
What to do here? Just revert to how it was before?

が居る  1,290,725   7.2% 
がいる 16,517,702  92.8% 

ている 715,720,121  99.9% 
て居る     450,959   0.1%

not that rare for senses 1 and 2.

Google N-gram Corpus Counts
╭─ーー─┬─────────────┬───────╮
│ 居る │   5.321.828 │  0.7% │[rK]
│ いる │ 804.530.319 │ 99.3% │
╰─ーー─┴─────────────┴───────╯

[ichi1] tag would need to be removed

@@ -5,0 +6 @@
+<ke_inf>&rK;</ke_inf>

@@ -15 +15 @@
-<xref type="see" seq="1296400">在る・1</xref>
+<xref type="see" seq="1296400">ある・1</xref>

Reverting.

@@ -17 +17 @@
-<gloss>to be</gloss>
+<gloss>to be (of animate objects)</gloss>

An edit like this needs to include detailed reasons.

@@ -17 +17 @@
-<gloss>to be (of animate objects)</gloss>
+<gloss>to be</gloss>

@@ -32 +32 @@
-<gloss>to have been ...-ing)</gloss>
+<gloss>to have been ...-ing</gloss>

So would I.

@@ -30 +30 @@
-<s_inf>after the -te form of a verb; indicates continuing action or state; the い is sometimes dropped</s_inf>
+<s_inf>after the -te form of a verb; indicates continuing action or state</s_inf>

I'd be happy to remove "the い is sometimes dropped" from the note.

@@ -30,2 +30,3 @@
-<s_inf>after the -te form of a verb, the い is sometimes dropped</s_inf>
-<gloss>verb indicating continuing action or state (i.e. to be ..ing, to have been ..ing)</gloss>
+<s_inf>after the -te form of a verb; indicates continuing action or state; the い is sometimes dropped</s_inf>
+<gloss>to be ...-ing</gloss>
+<gloss>to have been ...-ing)</gloss>

Move note about appearance "after the -te form of a verb" from 
gloss to actually a note.

@@ -30,2 +30,2 @@
-<s_inf>the い is sometimes dropped</s_inf>
-<gloss>(after the -te form of a verb) verb indicating continuing action or state (i.e. to be ..ing, to have been ..ing)</gloss>
+<s_inf>after the -te form of a verb, the い is sometimes dropped</s_inf>
+<gloss>verb indicating continuing action or state (i.e. to be ..ing, to have been ..ing)</gloss>

@@ -22,0 +23 @@
+<misc>&uk;</misc>
@@ -27,0 +29 @@
+<misc>&uk;</misc>

@@ -15,1 +15,1 @@
-<xref type="see" seq="1296400">在る</xref>
+<xref type="see" seq="1296400">在る・1</xref>

We decided on a 1M n-grams threshold for spec1. 解る and 判る meet this easily (when inflected forms are included).
But I don't think we can adequately explain the nuances of these forms in a short usage note.

@@ -11,0 +12 @@
+<ke_pri>spec1</ke_pri>
@@ -14,0 +16 @@
+<ke_pri>spec1</ke_pri>
@@ -29 +30,0 @@
-<s_inf>解る is used for comprehending; 判る is used for distinguishing or judging</s_inf>

わかっ	14795258	55.8%
分かっ	9635157	36.3%
解っ	1114047	4.2%
判っ	966361	3.6%

While that information about 解る and 判る is interesting, it's also rather marginal. GG5 and ルミナス don't even show those kanji forms and the kokugos downplay them. (I'm dropping the "spec1" tags).
I suspect it's probably adding some unnecessary clutter to the entry.

@@ -12 +11,0 @@
-<ke_pri>spec1</ke_pri>
@@ -16 +14,0 @@
-<ke_pri>spec1</ke_pri>
@@ -26 +23,0 @@
-<re_pri>spec1</re_pri>

* meikyo: 「判」は物事がそれと判別・判断される意で、「違い［善悪］が判る」「先のことは判らない」などと、「解」は理解できる意で、「英語が解る」「気持ちはよく解る」などと好まれるが、今は一般に「分」。「文学が 解る／判る」では、前者は単に理解する意だが、後者は傑作・駄作などをよく区別して理解する意となる。
* https://japanese.stackexchange.com/a/117:
    分かる: understand, know (a concept, a fact, etc)
    判る: (think 判断できる) judge, discriminate (between right and wrong, between two similar things, etc)
    解る: (think 理解できる) comprehend (a concept, an answer, etc)
* https://hinative.com/questions/995550
    分かる　is a very common term for the word "understand" (the fact, the concept etc...)
    判る　if you will look at the kanji you will think the word 判断できる which means judgement, discriminate, knowing the right from wrong, knowing between two similar things,
    解る just think the word 理解できる which means to comprehend like knowing the answer of the problem etc...

@@ -31,0 +32 @@
+<s_inf>解る is used for comprehending; 判る is used for distinguishing or judging</s_inf>

@@ -49,0 +50,5 @@
+<sense>
+<pos>&int;</pos>
+<gloss>I know!</gloss>
+<gloss>I think so too!</gloss>
+</sense>

G n-grams:
わかる	21853436
分かる	11513812
解る	 1077609
判る	  948510
分る	  339759
-----
わかっ	14795258
分かっ	9635157
解っ	1114047
判っ	 966361
分っ	 346395

gg5 uses kana in the examples

@@ -30,0 +31 @@
+<misc>&uk;</misc>
@@ -40,0 +42 @@
+<misc>&uk;</misc>

GG5, etc. too.

@@ -31,3 +31,6 @@
-<gloss>to be understood</gloss>
-<gloss>to be comprehended</gloss>
-<gloss>to be grasped</gloss>
+<gloss>to understand</gloss>
+<gloss>to comprehend</gloss>
+<gloss>to grasp</gloss>
+<gloss>to see</gloss>
+<gloss>to get</gloss>
+<gloss>to follow</gloss>

新和英中辞典 and prog have "(to) understand" rather than "to be 
understood" ..?

OK, I agree with that analysis.
スーパー大辞林 3.0 has the following explanation of the difference between 分かれる and 別れる:
［表記］ わかれる （ 分 ・別 ）
「分かれる」は“一つのものが別々のいくつかになる。違いが生じる”の意。「タクシー三台に分かれて乗る」「道が分かれる」「この作品については評価が分かれる」
「別れる」は“離れて去る”の意。「駅で友達と別れる」「家族と別れて暮らす」「夫婦が別れる」

-kojien has
わか・る【分かる・別る・判る・解る】
-daijs has
わか・る【別る】
although it specifies that it's actually the archaic form of 別れる.

-kojien doesn't make the distinction because it combines わか・れる【分かれる・別れる】, but since ~we~ keep our entries for 分かれる and 別れる separate, i don't think 別る should be added here.  also, it also can't be given its own entry because it's a v2

according to daijs, 分る is not io

@@ -19,2 +19,1 @@
-<keb>別る</keb>
-<ke_inf>&io;</ke_inf>
+<keb>分る</keb>

Eh? The only 別る I can find is あかる, which is not the same at all. Anyone else see a justification for this?

@@ -19,1 +19,1 @@
-<keb>分る</keb>
+<keb>別る</keb>

gg5

--> noun glosses

@@ -31,2 +31,2 @@
-<gloss>on hand</gloss>
-<gloss>ready</gloss>
+<gloss>what one has on hand</gloss>
+<gloss>what is ready</gloss>

I'll demote the ichi1 to ichi2.

@@ -10,1 +10,1 @@
-<ke_pri>ichi1</ke_pri>
+<ke_pri>ichi2</ke_pri>
@@ -23,1 +23,1 @@
-<re_pri>ichi1</re_pri>
+<re_pri>ichi2</re_pri>

hits

Hiragana-only gets more hits than any single surface form including kanji (though added together, those forms outnumber hiragana-only).
有り合せ only gets 18k hits, which doesn't really seem worthy of Priority. Suggesting removal of [ichi1] from that, and giving [spec1] to あり合わせ, which gets more hits than 有り合わせ.

@@ -4,0 +4,4 @@
+<k_ele>
+<keb>あり合わせ</keb>
+<ke_pri>spec1</ke_pri>
+</k_ele>
@@ -9,0 +13,3 @@
+<keb>あり合せ</keb>
+</k_ele>
+<k_ele>
@@ -10,1 +17,3 @@
-<ke_pri>ichi1</ke_pri>
+</k_ele>
+<k_ele>
+<keb>有りあわせ</keb>
@@ -15,0 +24,1 @@
+<re_pri>spec1</re_pri>
@@ -19,0 +29,1 @@
+<misc>&uk;</misc>

Give evocative literal translation, marked as [lit]

@@ -13,0 +13,1 @@
+<gloss g_type="lit">(the part that) does not step on the ground</gloss>

有りえない	12,602

@@ -12,0 +13,4 @@
+<ke_inf>&sK;</ke_inf>
+</k_ele>
+<k_ele>
+<keb>有りえない</keb>

Google N-gram Corpus Counts
╭─ーーーーー─┬───────────┬───────╮
│ あり得ない │   845,662 │ 19.5% │
│ 有り得ない │   404,028 │  9.3% │ - drop [spec1]
│ 有得ない　 │     5,724 │  0.1% │ - sK
│ ありえない │ 3,088,065 │ 71.1% │
╰─ーーーーー─┴───────────┴───────╯

@@ -10 +9,0 @@
-<ke_pri>spec1</ke_pri>
@@ -13,0 +13 @@
+<ke_inf>&sK;</ke_inf>

daijs:  あるはずがない。ありそうもない。そうなる可能性がない。

I don't agree with "unlikely; improbable". It's stronger than that.

@@ -21 +21 @@
-<xref type="see" seq="2851621">あり得る・ありえる</xref>
+<xref type="see" seq="2851621">ありえる</xref>
@@ -24,2 +23,0 @@
-<gloss>unlikely</gloss>
-<gloss>improbable</gloss>

@@ -21,3 +21 @@
-<xref type="see" seq="1540970">有り得る</xref>
-<xref type="see" seq="1540970">有り得る</xref>
-<xref type="see" seq="1540970">有り得る</xref>
+<xref type="see" seq="2851621">あり得る・ありえる</xref>

@@ -21 +21 @@
-<xref type="see" seq="1540970">有り得る・ありうる</xref>
+<xref type="see" seq="1540970">有り得る</xref>

I'm not sure about two either.

https://news.yahoo.co.jp/articles/00198485f37585914f759a70fcc49fca4ceccbba?page=2
「衆院選の真っ最中に党の重鎮である山崎氏が大阪10区にまで入って相手の辻元氏を応援なんてありえない」

not sure if really 2 senses, but looks ok to me.

@@ -25,0 +26,7 @@
+</sense>
+<sense>
+<pos>&adj-i;</pos>
+<misc>&uk;</misc>
+<gloss>unthinkable</gloss>
+<gloss>ridiculous</gloss>
+<gloss>absurd</gloss>

ルミナス (for extra glosses)

700k hits

@@ -12,0 +12,3 @@
+<k_ele>
+<keb>有得ない</keb>
+</k_ele>
@@ -21,0 +24,2 @@
+<gloss>unlikely</gloss>
+<gloss>improbable</gloss>

@@ -18,0 +18,2 @@
+<xref type="see" seq="1540970">有り得る・ありうる</xref>
+<xref type="see" seq="1540970">有り得る・ありうる</xref>

Entry created

eij
ordered after bing results (17k, 11k, 10k (ca))

@@ -4,0 +4,6 @@
+<k_ele>
+<keb>触らぬ神に祟りなし</keb>
+</k_ele>
+<k_ele>
+<keb>触らぬ神にたたりなし</keb>
+</k_ele>
@@ -13,1 +19,2 @@
-<gloss>Let sleeping dogs lie</gloss>
+<gloss>let sleeping dogs lie</gloss>
+<gloss>if it ain't broke, don't fix it</gloss>

Entry created

koj, daij, gg5, eij

Well, my approach to dealing with the additive versus sequential distinction was to add that other sense. Personally, I feel that if distinguishing those two is too much, then we cannot really justify all the other senses, which arguably bear a less important contrast. And if that is so... we might as well just reduce them all to a single sense.
As to the terminology, I am not sure if "additive relation" is immediately intuitive, but probably should be fine given that it is accompanied by a translation.
In any case, my input on how to word this can be found under my older proposal - just scroll down a bit and you will find an obscenely large change log under an edit by Mr. Breen, which was that proposal being reverted. Sadness, sadness.

Now; the reason why I made で a reading and the others not, is because it makes the entry and its notes shorter. 
て and で are the endings of all て forms; so - in theory - one should be able to identify all instances by those endings. Thence my note says "becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ" instead of "becomes いで after 五段 verbs ending in ぐ; becomes んで after 五段 verbs ending in ぬ、ぶ or む". 
Consider also that adding って means that we are missing んで, いで and いて in the readings and notes. Needless to say, the readings might be fine but such notes would end up covering a metric mile - which I am not necessarily against, but I am not the one you will need to convince anyway so...

Regarding your first point, I had that in mind, but I hadn't yet thought of a way to deal with that. I feel like having two senses may be a bit much. 
Perhaps having "indicates sequential or additive relation" works? That should be somewhat clear then with the glosses.
Or maybe "indicates simple conjunction or sequential relation"?

The "forming compound verbs" also checks out.

＞ I also have reverted the term 'subsidiary' to 'auxiliary' as I think the latter is more common and more likely to be immediately understood (plus, it is 1 letter shorter, which will surely make all the difference)
   ＞ I can live with that, it's indeed more common and easier to understand. I prefer to make the distinction (as do the references), but it's not typically done in JMdict AFAICS. 
      Then that xref can also go again

Regarding the attachment note, I agree it's probably better to leave it out here. It'd be more of a matter of formality rather than practicability.
＞ (maybe do something on the 音便形 page and cross reference it here?)
   ＞ I was thinking along these lines as well, but that'd most likely turn out as a very long and windy explanatory gloss.
      I believe this is fine

Since we are at 音便 already, this entry also needs to mention って, as in 知りて ＞ 知って, 立ちて ＞ 立って and 買ゐて＞買って.
I propose just adding that to the note. 

I have added って as a surface form

@@ -9,0 +10,3 @@
+<r_ele>
+<reb>って</reb>
+</r_ele>
@@ -13 +16 @@
-<s_inf>after the continuative form; indicates sequential relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>indicates sequential relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>
@@ -15 +18 @@
-<gloss>then</gloss>
+<gloss>(and) then</gloss>
@@ -20 +23 @@
-<s_inf>after the continuative form; indicates causal relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>indicates causal relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>
@@ -28 +31 @@
-<s_inf>after the continuative form; indicates instrumental relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>indicates instrumental relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>
@@ -35 +38 @@
-<s_inf>after the continuative form; indicates contrastive relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>indicates contrastive relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>
@@ -42,2 +45 @@
-<xref type="see" seq="1514640">補助動詞</xref>
-<s_inf>after the continuative form; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>
@@ -44,0 +47 @@
+<gloss>particle used to form compound verbs (as in やって来る)</gloss>
@@ -49 +52 @@
-<s_inf>at sentence end; indicates request or light imperative; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>at sentence end; indicates request or light imperative; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ; って after 五段 verbs ending in る, う or つ</s_inf>

Regarding [1]: The conjunction is not necessarily always sequential, it often its, and て has a tendency to chosen over the 連用形 if the relation between the conjuncts is sequential, but it is not restricted to those uses.
The most obvious case is with adjectives: 「黒くて高い」can be freely rephrased as 「高くて黒い」since no sequence is implied. This is why my older proposal had 7 senses as opposed to the current 6 - it had an extra one for the bare, non-sequential conjunction, which was [1].


Regarding compound verbs： I am referring to cases wherein verb + て + verb is analysed as just one verb. This use, like pretty much all others, is rooted in sense 1. Albeit rarer than the 連用形 compounds, it works under the same principles.
For example, 遣って来る, despite clearly being made from やる＋て＋くる、is given its own entry in Daijirin, Daijisen, Shinmeikai and Koujien. In all of these cases, it is listed as a カ変, rather than the expected 連語. There are more entries like this such as 取って代わる.
Perhaps an even better example is 打って変る, whose meaning is not really the sum of its parts.

Regarding the [conj] particle: As briefly mentioned above, pretty much all senses of て are rooted in its role as a conjunctive (which itself is rooted in its origin as the 連用形 of the ancient perfective auxiliary つ), with the exception being senses [5] and [6] of the current proposal wherein it cannot really be analysed as a conjunction - which is why I did not put in the tag and am thus removing it from [5]. 
I also have reverted the term 'subsidiary' to 'auxiliary' as I think the latter is more common and more likely to be immediately understood (plus, it is 1 letter shorter, which will surely make all the difference).

And lastly, stem connection: While the stem for the て form is regarded as being the 連用形 due to the etymological origins of the 音便形 as a sound change of the 連用形 + て, in practice the 音便形 is its own inflection.
As such, writing either 連用形, continuative form, or -masu stem, will not allow a reader to correctly deduce the inflection - instead they would end up producing an inflection right out of Old Japanese, such as 書きて instead of 書いて, etc. - this is why both foreign Japanese textbooks and native Japanese textbooks aimed at foreigners almost unanimously teach it as "て-form" or て形、 as an inflection itself.
I am not sure if there is a good way to convey that in this dictionary (maybe do something on the 音便形 page and cross reference it here?) as this is complicated by the fact that the "音便 stem" involves alteration of the verbal root.
I.e. Kak-, the root 書く, must lose the final k, receive the left-over -i from the 連用形, turn into Ka-i- and only then connect to て - which is unlike the other stems that are all simply concatenated onto the root: Kak-a/i/e/o/u.
Either way, I would drop that part of the note.

@@ -42 +41,0 @@
-<pos>&conj;</pos>
@@ -45 +44 @@
-<gloss>particle used to connect to subsidiary verbs (as in て-いる, て-いく, etc.)</gloss>
+<gloss>particle used to connect to auxiliary verbs (as in て-いる, て-いく, etc.)</gloss>

(See also Shinsen's entry for this, I can't add it here right now due to the issue described below)

Progressive 4：
－て

1〔事を終えて次に連続する意〕and
私は朝食を済ませて出掛けた
I had my breakfast and went out.／After finishing my breakfast, I went out.

2〔原因・理由を表す〕
重くて持ち上がらない
It's so heavy (that) I can't lift it.／It's too heavy for me to lift.
うれしくて飛び上がった
I jumped for joy.

3〔手段・方法を表す〕
フランス語を教えて生計を立てている
She earns her living by giving lessons in French.

4〔…のに〕
知っていて黙っていた
He knew about it, but he did not tell me.／Although he knew, he didn't tell me.

5〔軽い命令を表す〕
やめて
Stop it.

ーーーーーーーーーーーーーーーーーーーー

Firstly, I intended to submit my draft for this entry in this comment box at the bottom instead of outright changing all you have proposed (which I do agree with). 
However, when then attempting to submit that I get the error "The requested URL's length exceeds the capacity limit for this server.", which I haven't encountered so far.

ーーーーーーーーーーーーーーーーーーーー
＞ As I am editing this, [3] Is being shown as "[prt][col]" in the preview instead of "[prt,col]" despite having been formatted like the latter [...]
I reckon that is because they belong to different tag classes. [prt] belongs to the POS tags, while [col] is a misc tag.

＞ I fear that might lead a reader to believe that said note applies only to [1], which would be misleading.
I fear so too
ーーーーーーーーーーーーーーーーーーーー
Being a (relatively) new contributor to JMdict, I've never really looked at this entry, but I absolutely agree that the former version is very much insufficient. 
ーーーーーーーーーーーーーーーーーーーー
Also, I see that you've brought up the point about 接続助詞 below. The good news is that a [conj-prt] tag will most likely arrive in the near future, as I've raised that topic in the github issues.

Could you elaborate on sense 2 "particle used to form compound verbs"? I don't quite see that in references, since it's usually the bare 連用形 used for this purpose.

Stem attachment information tends to be included. However, adding the typical "after the -masu stem of a verb" doesn't suffice here. Perhaps "After the continuative form" is appropriate.
   連用形 and the likes are typically not used in sense notes in JMdict. Note that "continuative form" is searchable in JMdict and leads directly to the entry for 連用形.

The references I consulted (Shinsen, Sankoku, Daijirin) class the current sense 2 as "接続助詞" as well, so I added [conj] for now.

I see that the support has previously been lacking, but I believe that having separate senses for the different uses of the conjunctive aspect is necessary. 
   Compare shinsen above, which is quite neat, but not achievable in JMdict in that fashion.  
   I would argue that the entries for the most used "words" (items would be better), deserve the most care, i.e. specificity. And that does take time in some cases
ーーーーーーーーーーーーーーーーーーーー
Thus I propose something along the lines of this (subject to refinement)

@@ -13 +13 @@
-<s_inf>joins two sentences; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>after the continuative form; indicates sequential relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
@@ -14,0 +15 @@
+<gloss>then</gloss>
@@ -18,2 +19,27 @@
-<s_inf>becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
-<gloss>particle used to connect auxiliary verbs or to form compound verbs</gloss>
+<pos>&conj;</pos>
+<s_inf>after the continuative form; indicates causal relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>because (of)</gloss>
+<gloss>so</gloss>
+<gloss>since</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>after the continuative form; indicates instrumental relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>through (that)</gloss>
+<gloss>by</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>after the continuative form; indicates contrastive relation; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>although</gloss>
+<gloss>but</gloss>
+<gloss>(and) yet</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<xref type="see" seq="1514640">補助動詞</xref>
+<s_inf>after the continuative form; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>particle used to connect to subsidiary verbs (as in て-いる, て-いく, etc.)</gloss>
@@ -24 +50 @@
-<s_inf>at sentence end; used to make a request; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<s_inf>at sentence end; indicates request or light imperative; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>

Well, I did say that they were there to be picked out, and that only [1], [6] and [7] were strictly necessary. I still stand by what I said that this entry cannot stay like this, so, let us try that.
Repeating the note on で in every sense might be unneeded, maybe remove it from [2] and [3], leaving it only on [1]. However, I fear that might lead a reader to believe that said note applies only to [1], which would be misleading. Really wish we had notes that went over the entire glossary...

As I am editing this, [3] Is being shown as "[prt][col]" in the preview instead of "[prt,col]" despite having been formatted like the latter, hopefully it will actually be like that when I submit it, but if it does not, I do not know why it is like this.

@@ -6,0 +7,3 @@
+<r_ele>
+<reb>で</reb>
+</r_ele>
@@ -8,3 +11,15 @@
-<pos>&aux;</pos>
-<s_inf>で after certain verb forms</s_inf>
-<gloss>indicates continuing action</gloss>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>joins two sentences; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>and</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<s_inf>becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>particle used to connect auxiliary verbs or to form compound verbs</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<misc>&col;</misc>
+<s_inf>at sentence end; used to make a request; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>please (do)</gloss>

This long edit obviously is not getting any support. I'm switching it back to the predecessor, but leaving Non's suggestions visible. I found it to be far to detailed and complex to be a useful entry.

@@ -7,3 +6,0 @@
-<r_ele>
-<reb>で</reb>
-</r_ele>
@@ -11,39 +8,6 @@
-<pos>&prt;</pos>
-<pos>&conj;</pos>
-<s_inf>joins two sentences; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
-<gloss>and</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<pos>&conj;</pos>
-<s_inf>connects sentences in order of occurrence</s_inf>
-<gloss>and (then)</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<pos>&conj;</pos>
-<s_inf>connects sentences in causal relation</s_inf>
-<gloss>and (because of that)</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<pos>&conj;</pos>
-<s_inf>connects sentences in instrumental relation; preceding sentence is means by which following sentence takes place</s_inf>
-<gloss>and (through that)...</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<pos>&conj;</pos>
-<s_inf>forms contrastive conjunction</s_inf>
-<gloss>but...</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<xref type="see" seq="1577980">いる・3</xref>
-<gloss g_type="expl">used to connect to auxiliary verbs or to form compound verbs</gloss>
-</sense>
-<sense>
-<pos>&prt;</pos>
-<misc>&col;</misc>
-<s_inf>at sentence end</s_inf>
-<gloss g_type="expl">used to make a request</gloss>
+<pos>&aux;</pos>
+<xref type="see" seq="2028980">で・5</xref>
+<xref type="see" seq="2028980">で・5</xref>
+<xref type="see" seq="2028980">で・5</xref>
+<s_inf>で after certain verb forms</s_inf>
+<gloss>indicates continuing action</gloss>

I made somewhat of a revision of my proposal. As this has been drifting for months and nobody seems to want to touch it, I would like to highlight that I was not saying all these senses are needed, I made them in bulk so that an editor could pick them out if needed - the entry could be reduced to only [1], [6] and [7].

At any rate, I think the earlier versions of this unapproved entry should be closed, at least.

@@ -13,2 +13,2 @@
-<s_inf>forms plain conjunction; 1-7 become ～で after godan verbs ending in ぐ, む, ぶ or ぬ</s_inf>
-<gloss>and...</gloss>
+<s_inf>joins two sentences; becomes で after 五段 verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>and</gloss>
@@ -20 +20 @@
-<gloss>and then...</gloss>
+<gloss>and (then)</gloss>
@@ -26 +26 @@
-<gloss>and because of that...</gloss>
+<gloss>and (because of that)</gloss>
@@ -31,2 +31,2 @@
-<s_inf>connects sentences in instrumental relation</s_inf>
-<gloss>and through that... (preceding sentence is the means by which following sentence takes place)</gloss>
+<s_inf>connects sentences in instrumental relation; preceding sentence is means by which following sentence takes place</s_inf>
+<gloss>and (through that)...</gloss>
@@ -43 +43 @@
-<gloss g_type="expl">used to connect verbs to auxiliary verbs or to form compound verbs</gloss>
+<gloss g_type="expl">used to connect to auxiliary verbs or to form compound verbs</gloss>
@@ -48 +48 @@
-<s_inf>used at sentence end</s_inf>
+<s_inf>at sentence end</s_inf>

Proposing adding で to readings to go along with the proposal of deleting the sense of the cross-reference leading here.

@@ -5,0 +6,3 @@
+</r_ele>
+<r_ele>
+<reb>で</reb>

daijr, daijs

Changed [aux] to [prt] as it is classed as a 接続助詞. Completely changing glossary as "Indicates continuing action" will likely lead anyone learning of this word through this entry to think of a continuous verb rather than a conjunction.
Senses are based on the references listed, mainly daijr, they are not absolutely exhaustive but these should be the main ones. I opted to make [1] what it is as most other senses can be imposed upon it by context.

If accepted, I would like to suggest deleting the 5th sense of the entry on ～で that leads here: it is same word and the changes to notes in this entry should make it unnecessary.

@@ -8,4 +8,39 @@
-<pos>&aux;</pos>
-<xref type="see" seq="2028980">で・5</xref>
-<s_inf>で after certain verb forms</s_inf>
-<gloss>indicates continuing action</gloss>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>forms plain conjunction; 1-7 become ～で after godan verbs ending in ぐ, む, ぶ or ぬ</s_inf>
+<gloss>and...</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>connects sentences in order of occurrence</s_inf>
+<gloss>and then...</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>connects sentences in causal relation</s_inf>
+<gloss>and because of that...</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>connects sentences in instrumental relation</s_inf>
+<gloss>and through that... (preceding sentence is the means by which following sentence takes place)</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<pos>&conj;</pos>
+<s_inf>forms contrastive conjunction</s_inf>
+<gloss>but...</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<xref type="see" seq="1577980">いる・3</xref>
+<gloss g_type="expl">used to connect verbs to auxiliary verbs or to form compound verbs</gloss>
+</sense>
+<sense>
+<pos>&prt;</pos>
+<misc>&col;</misc>
+<s_inf>used at sentence end</s_inf>
+<gloss g_type="expl">used to make a request</gloss>

@@ -9,1 +9,1 @@
-<xref type="see" seq="2028980">で</xref>
+<xref type="see" seq="2028980">で・5</xref>

I wonder about the "aux". UniDic has "助詞,接続助詞" for this. I guess "aux" covers a multitude of sins.

@@ -9,0 +9,2 @@
+<xref type="see" seq="2028980">で</xref>
+<xref type="see" seq="2028980">で</xref>
@@ -10,1 +12,1 @@
-<gloss>used to indicate continuing action</gloss>
+<gloss>indicates continuing action</gloss>

actually, come to think of it, れる and られる could be merged with a note.  they're already sufficiently well-documented.

they have to be merged because one is merely an alternate pronunciation of the other that occurs with certain verb forms. れる and られる are clearly more divergent than て and で.

in addition, there is already an existing entry for で, and we try to avoid having entries with identical headword/reading combos.

@@ -7,3 +7,0 @@
-<r_ele>
-<reb>で</reb>
-</r_ele>
@@ -12,0 +9,1 @@
+<s_inf>で after certain verb forms</s_inf>

If these 2 needs to be merged as RENE suggested, I have no 
idea why れる・られる　or せる・させる are not merged.

All these require to follow complicated conjugation rule but 
the same word for each.

Criteria for an entity is a bit unclear.

@@ -6,0 +6,3 @@
+</r_ele>
+<r_ele>
+<reb>で</reb>

http://ja.wikipedia.org/wiki/助詞
http://ja.wiktionary.org/wiki/て

used with -masu stem verbs except for v5g, v5n, v5b, v5m

Daijirin, Ｓhinmeikai.
Daijirin entry for ず*:
『（助動）
1 現代語の打ち消しの助動詞「ぬ」の連用形。→ぬ（助動）
2 ず（ざら）・ず（ざり）・ず・ぬ（ざる）・ね（ざれ）・ざれ
古語の打ち消しの助動詞。用言およびある種の助動詞の未然形に接続する。打ち消しの意を表す。ない。ぬ。』
The（助動）at the top applies to the entire entry.
*The numbers are mine; the numerals used by Daijirin for its senses are images and cannot be copied so I had to replace them.

"My idea was that sense one is not something that inflects, it's the 連用形 of ぬ *singled out* to have its separate sense. 
Thus it does not inflect, and is basically used as a 未然形-bound 接続助詞."
ずに, ずんば and entirety of the ざり paradigm found dead in the streets of the 平安 period.
Jokes aside, you can take that approach - but this is entirely a matter of philosophy wherein ぬ is regarded as the flexible original and all other forms as immutable variants.
As a rule, dictionary forms are chosen primarily by pragmatism and not by any morphological criteria i.e. ぬ and ず are both inflections of the same lexeme, which we call ぬ by convention - but that does not mean that ぬ turns into ず or vice-versa, or that they do not.
You *can* posit that only ぬ inflects, but it is very hard to actually prove it.

"The kokugos that perform this same split do not assign a PoS tag to the 連用形 ず sense."
Well, the kokugos do not have the constraint that they need a PoS tag or else they get a syntax error.
In any case, Daijirin and Shinmeikai have both senses under 助動詞; since we need a tag I still think [aux-v] is best.

"And I do remember seeing that style of note in another entry (it might have been Non there too, I don't quite recall)"
I do think the only instances of a ~ between a stem and suffix are the ones I edited (and I think they are all the せ～ 未然形 + suffix, courtesy of "-nai stem") but the practice of prefixing a ~ to suffixes, etc. in notes predates me by several years. 
I first used the intermediary ~ in either the (ら)れる or (さ)せる entry, which already had a ~ before I added the stem.  

"This perhaps isn't very relevant to the ず discussion in particular, but I was under the assumption that there is syntactical difference between these two. Is there not?"
That depends on the level of morphosyntactical analysis and the school of thought/philosophy being employed to execute it, as they define where we want our morpheme boundaries to be and how the syntax trees are structured.
If we define our boundaries to be something relatively simple such as semantic content and grammatical function, both are identical in that they consist of a base content verb + a suffix adding grammatical information: 
せぬ = se-nu = do-NEG.NPS
している = shite-iru = do-CONT.NPS 
The morphological structure may then be reflected in the syntax tree (since the stems せ and し, as well as て, have neither semantic content nor overt grammatical function in their current use, they are ignored by this analysis).

Finer levels may take every single root, meaningless stem and particle, or even concepts not overtly realised into their glosses:
せぬ = s-e-n-u-∅ = do-LK-NEG-NPS-PRED 
している = s-i-て-i-る-∅ = do-LK-PTCL-CONT-NPS-PRED
At this point they are no longer identical, but the very categories of 助動詞 and 補助動詞 are no longer consistent within themselves either (if they ever were to begin with).
The truth is that conventional grammatical categories are a compromise between accuracy and practicity: are 助動詞 and 補助動詞 the same? At a certain degree of compromise, yes; at another, no. And as you play around with the north of your compass, they may grow as far apart from themselves as they do from each other. 

CONT=Continuous aspect
LK= Interfix, here the stem's vowel.
NEG=Negative
NPS=Non-past
PTCL=Particle
PRED=Predicative

Meikyo:
ず
［打ち消しの助動詞「ぬ」の連用形］
❶打ち消しの意を伴って、文を中止したり、副詞的修飾をしたりする。…ないで。
[second sense excluded here]

Sankoku:
ず
◇
口語の否定の助動詞「ぬ」の連用形。
「何も買わ━、帰った」
[second sense excluded here]

↑
Oukoku and Smk do the same

.

＞ Plus, it is important to distinguish between suffixes that inflect from inflectional suffixes: most 助詞 like the case-marking particles, the prohibitive な, etc. are also inflectional suffixes. 
That's true, I didn't phrase that clearly enough. I meant "suffixes that inflect".
ーーー

＞A 補助動詞 is functionally identical to a 助動詞, the difference being lexicography and etymology: [...]
This perhaps isn't very relevant to the ず discussion in particular, but I was under the assumption that there is syntactical difference between these two. Is there not?
ーーー

＞The [aux] tag by itself is a tad nebulous, it is probably best reserved for such words that even standard Kokugos fail to classify or disagree as to the category.
With [aux-v] it is at least clear that it is an auxiliary that inflects.

My idea was that sense one is not something that inflects, it's the 連用形 of ぬ *singled out* to have its separate sense. 
Thus it does not inflect, and is basically used as a 未然形-bound 接続助詞.
The kokugos that perform this same split do not assign a PoS tag to the 連用形 ず sense.
See the quoted references (I saw that daijr does have 〘助動〙, but I'm pretty sure that's because daijr doesn't use split PoS tags like this in the same entry? I may be wrong though)
That's why I suggested not using [aux-v] here, *if* we define [aux-v] as something that inflects.
ーーー

As for the discussion about the note format, I don't have any particular opinion on it.
However, the arguments brought forth by Non seem reasonable to me, after all we have a sufficiently explanatory entry for "サ変".
And I do remember seeing that style of note in another entry (it might have been Non there too, I don't quite recall)

"I don't like the use of サ変 in the note and I'm not sure why there's a tilde between せ and ず."
サ変 was chosen over suru-class or something of the sort as it is mildly shorter and refers to the conjugation class as a whole rather than する singularly.
While す・する and サ変 are largely synonymous since the former and its compounds very likely account for 99.9% of the conjugation class, there are very few verbs that occupy that 0.1% position wherein they have the suru-class paradigm, despite having no connection to the actual suru.
Off the top of my head I can recall 御座す(おはす), 坐す(います) and まらする；the former probably began as a either a 二段 or 四段 from 座す(わす) while the latter two were originally 四段 and 二段 respectively, and only later became サ変.

As for the tilde, it is there to separate the stem from the suffix, modelled after a handful of entries that have a tilde in front of an auxiliary, compound, set-phrase, etc. when it appears in a note.
Is it strictly necessary? Probably not.
But personal experience tells me that if something can be interpreted wrongly, it will. And so I reckon there will be at least one person who will read "after the -nai stem of a verb and as せず for する" as ず sprouting a せ in front of it and then attaching to する as しせず.
Cursed be the -nai stem...
All in all, I would rather keep it.

"I'm now very well aware that "auxiliary verb" is a rather problematic term for what are in essence inflectional suffixes. The overlap between "auxiliary verb" in indo-european languages and the term 助動詞 in 国語文法 is rather minimal. It would much rather be the 補助動詞 falling under that notion."
We are mixing terminology here;　"auxiliary verb" in the context of Japanese refers to either 助動詞 and/or both 補助動詞 and 助動詞, it has little to do with what the term may mean in other languages. 
Plus, it is important to distinguish between suffixes that inflect from inflectional suffixes: most 助詞 like the case-marking particles, the prohibitive な, etc. are also inflectional suffixes. 
助動詞 Are specifically inflectional suffixes that - like 動詞 - have their own inflectional paradigm, thence the name.
A 補助動詞 is functionally identical to a 助動詞, the difference being lexicography and etymology: 補助動詞 are bound morphemes considered to be a single lexeme with the unbound morpheme they derive from, a 助動詞 either has no known associated unbound morpheme (ぬ) or has one but has since evolved into its own lexeme (ます ultimately comes from 参る).

"It is probably best to reduce this just to [aux] for the time being."
The [aux] tag by itself is a tad nebulous, it is probably best reserved for such words that even standard Kokugos fail to classify or disagree as to the category.
With [aux-v] it is at least clear that it is an auxiliary that inflects.

What I forgot to mention was that if the things I suggested at the end of my reply have already been discussed sometime somewhere (e.g. in the mailing list), it would be nice if you could link them here/refer to them. Thanks

＊ https://web.archive.org/web/20021024101613/http://homepage3.nifty.com/taketoki/jodosiless.html
Rather long but very detailed.

＊ https://ja.wikipedia.org/wiki/助動詞_(言語学)
「国文法では、この助動詞を補助動詞と呼び[1]、「-た」や「-れる・-られる」などを「助動詞」と呼ぶ。言語学ではこれらは英語の -ing や -ed と同様に語尾や接尾辞と見なされる[2][3][4]。」

＊Nikkoku
助動詞：「日本語で付属語のうち、活用のあるもの。他の自立語（詞）、または自立語を含む連語に付属して、叙述の意義を補ったり、話手の判断の性質を表現したりする。助詞とともに、付属語、または辞と呼ばれる。動辞。」


.

The idea (or at least my idea) behind this entire string of edits was to have the singular form of 連用中止 ず have its separate sense. 
In that regard it could very well be thought of as a kind of 接続助詞 (though "conjunctive suffix" would fit better (I believe that a 接続助詞 tag would perhaps actually be usable here)) in my eyes. 
I have now realized that [aux-v] is problematic as well here, as they, by definition, inflect. Which this singled out 連用形 doesn't.


＞Someone on the mailing list requested that [aux-v] be removed from this entry, arguing that ず is not an auxiliary verb but an auxiliary suffix. I want to note here that the kokugos classify it as a 助動詞 (auxiliary verb), like ぬ.　Japanese auxiliary verbs are always 付属語 (dependent words).


I've also read that recent email exchange, and noticed that I actually took a different stance on the suffix aspect in my first edit here, but I have changed my mind since. 
I'm now very well aware that "auxiliary verb" is a rather problematic term for what are in essence inflectional suffixes. The overlap between "auxiliary verb" in indo-european languages and the term 助動詞 in 国語文法 is rather minimal. It would much rather be the 補助動詞 falling under that notion.
It's just that all of these "auxiliary verbs" are currently handled a bit different in jmdict it seems.

It is probably best to reduce this just to [aux] for the time being. And this is all I'm suggesting when it comes to changes, I have nothing to add on the other edits made to this entry.


I have been looking into part of speech categorization in Japanese quite a lot after your edit, and I can't say I've come to a conclusion yet (It just lead to some more confusion, which I'll have to resolve first).
But something that generally floated in my mind previously was the idea of (trying to) define what falls under these otherwise somewhat ominous jmdict tags of [suf], [aux], [aux-v], [aux-adj] (and maybe [cop]). I think that would benefit everyone involved. I'm not suggesting to do this here, but rather putting the general suggestion up for debate.

Another thing would maybe deciding on a standardized note format for the above tags.

I don't like the use of サ変 in the note and I'm not sure why there's a tilde between せ and ず. I've suggested a rewording of that part.
I don't think "negates verb and continues onto next sentence" is needed. We have "continuative form of ぬ" in the note.
I'm not comfortable with [conj] here. Although ず serves a conjunctive role, it's neither a 接続詞 nor a 接続助詞.
Someone on the mailing list requested that [aux-v] be removed from this entry, arguing that ず is not an auxiliary verb but an auxiliary suffix. I want to note here that the kokugos classify it as a 助動詞 (auxiliary verb), like ぬ.　Japanese auxiliary verbs are always 付属語 (dependent words).

@@ -9 +8,0 @@
-<pos>&conj;</pos>
@@ -12,6 +11,5 @@
-<s_inf>after the -nai stem of a verb and for サ変 as せ～ず (all senses); continuative form of ぬ</s_inf>
-<gloss>(does) not... and...</gloss>
-<gloss>not doing...</gloss>
-<gloss>not... so...</gloss>
-<gloss>without doing</gloss>
-<gloss>negates verb and continues onto next sentence</gloss>
+<s_inf>after the -nai stem of a verb and as せず for する (all senses); continuative form of ぬ</s_inf>
+<gloss>(does) not ..., and</gloss>
+<gloss>(does) not ..., so</gloss>
+<gloss>not (doing)</gloss>
+<gloss>without (doing)</gloss>

I'm not quite sure why I added the [form], it's not really in the references I quoted and I don't think ず 中止法 is [form].
The closest would be that daijisen quote.

I don't think it was ever mentioned in the discussion in the first place, so I assume there's no opposition to removing it anyway.

@@ -12 +11,0 @@
-<misc>&form;</misc>

Seems to work.

＞ Actually, I think I like this style (used on 良い) more. It's both natural and easily machine readable.
   ＞ Currently there's no dominant form. There are exactly three entries with such a note, て, よい and this entry now. 
      I would prefer it the way it was before, at the start separated by a semicolon (though that obviously doesn't work now since I have left "continuative form of ぬ" in the note)
      However, honestly, I don't think it's THAT necessary in an entry with "only" two senses. It really only becomes annoying for entries precisely like よい and て.
      That said, I won't make a fuss over it, it's also fine this way. 
      

＞ 二段; I would not worry too much about them. Not only would that mean we still need to add in 四段、ナ変、ラ変 and the ancient 未然形 of adjectives [...]
   ＞ This is also what I was getting at in my previous comment. I suggested generalizing from "verb class A, B, C, and as X to サ変" to "verbs, and as X to サ変". 
      And I would still be in favour of doing so

As for nonさん's revision:

Having "negates verb and continues onto next sentence" in the glosses certainly helps. 
However, though rather minor, I would prefer for "continuative form of ぬ" to stay in the notes.

As for the last gloss using just "verbs" (instead of generalizing or having "adjectives and verbs"), I don't think it's very significant, but I checked anyway out of curiosity.
(Excluding the debate on whether 形容詞 should be analyzed as static verbs)
Since constructions like 〇からず、 are possible I searched in NINJAL's SHC (昭和・平成書き言葉) Corpus for instances of "｛から｝｛ず｝｛、｝". 
It returned "only" 15 results, of which the ず is analyzed as belonging to 文語助動詞　ず instead of ぬ for some of those meager 15.
Though I assume (correct me in case I spout rubbish) カリ活用 was already a bit out of fashion at that time.
In any case I'm just throwing this information out there, perhaps it's relevant perhaps it isn't. 


Another minor change is, as it's already pretty difficult to keep a general consistency with these notes, to have "after the X form of X" instead of "connects to X stem of X".
Purely based on the fact that there are as of now 195 (maybe a few false hits) entries with notes using "after the", compared to 3 using "connects to" (2 without this entry).
Unless there are objections of course.



In the meantime I will attempt to master the art of sitting down and only then walking in, wish me luck.

@@ -13 +13 @@
-<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず (all senses); continuative form of ぬ acting as a negative conjunction</s_inf>
+<s_inf>after the -nai stem of a verb and for サ変 as せ～ず (all senses); continuative form of ぬ</s_inf>
@@ -15 +15 @@
-<gloss>not doing</gloss>
+<gloss>not doing...</gloss>
@@ -17,0 +18 @@
+<gloss>negates verb and continues onto next sentence</gloss>

Actually, I think I like this style (used on 良い) more. It's both natural and easily machine readable.

@@ -13 +13 @@
-<s_inf>[all senses]: connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず; continuative form of ぬ acting as a negative conjunction</s_inf>
+<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず (all senses); continuative form of ぬ acting as a negative conjunction</s_inf>

> I agree that the repeated note is unsightly, this is again the problem wherein we have no means to distinguish a sense-specific note from an entry-wide one.

This is similar to the fix Jim added to the て entry (2345109). The only thing I changed was adding brackets to make it more programmatically readable (e.g., via regex). Obviously the most appropriate answer would be to add a new element to denote entry-wide notes, however, this works for now, and keeps compatibility with existing tools.

@@ -13 +13 @@
-<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず; continuative form of ぬ acting as a negative conjunction</s_inf>
+<s_inf>[all senses]: connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず; continuative form of ぬ acting as a negative conjunction</s_inf>
@@ -23 +22,0 @@
-<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず</s_inf>

In response to McBroom:
I agree that the repeated note is unsightly, this is again the problem wherein we have no means to distinguish a sense-specific note from an entry-wide one. 

二段; I would not worry too much about them. Not only would that mean we still need to add in 四段、ナ変、ラ変 and the ancient 未然形 of adjectives that is so old not even Old Japanese used it, but also the point of the note is more so to clarify to readers unfamiliar with ず that the connection to サ変 is せず and never しず, as would be otherwise implied by "-nai stem". 
In any case, I will leave an alternative entry without the repeating note at the bottom of this commentary.

In response to Sombrero:
The universality of the relations that can be expressed by a verbal coordinator (here ず) has less to do with the coordinator and more to do with the many possible relations between the propositions being joined, which are contextually inferred but may have downstream semantic and syntactical effects, such as the autonomously advancing reference time and subordinating-like set sentence order imposed on this example:
"John walked in and sat down"
In theory, if this is just a coordination, I should also be able to say "John sat down and walked in" to mean the same, but I cannot - we infer a sequence which that would violate.
And undeniably, few have ever mastered the art of sitting down and only then walking in.

And all this translates into: I do not think we should be too worried about listing every possible relation between the conjuncts since it is not ず providing them anyway. Just the basic coordination and maybe a few of the more common inferred relations such as 'without' are likely enough.

Lastly, yes - ず is hard to gloss. This is so because we have no 1:1 correspondent to it in English, even less so than normally.
ず negates and coordinates. We have things that negate and things that coordinate, but not both (actually we have 'nor', but it can only join two negatives whereas ず is not so picky).
This is perfectly illustrated by the translations you chose at the end, both of them contain at least one conjunctive and one negative that come from ず, but in both cases they need to be separated and put into different positions in the phrase, which gives the impression that ず is lost in translation but in reality it is being spliced.
Thence both you and I feel this need to leave an explanation somewhere, even the most literal translations such as 'not... and...' seem fuzzy as they do not clearly confer the underlying knowledge that 'not' stands for the negation of one clause and 'and' stands for conjunction with the following one.

Below is the attempt at a revision without the repeated notes:
[1][aux-v,conj][form]
  [note="all senses: connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず"]
  (does) not... and...; not... so...; without (doing); continuative form of ぬ, negates verb and continues onto next sentence
  [see=2441300・ぬ[1]]
  [see=2576210・ずに[1]]
[2][aux-v][arch]
   not
  [see=2257550・ない[1]]

＞ In short, it is the サ変 -nai stem problem again - I once again would like to address it, if nothing else.
   The note might seem overly long but ironically is shorter than other alternatives I have tried as the kanji save us a handful of characters.
   ＞I think it's fine to have a note for サ変, 未然形 せ is certainly not encountered early on by learners, better to have it listed

＞ Also, shouldn’t the archaic form be noted as connecting to archaic nidan verbs as well (e.g., 捨つ->捨てず)
   ＞ Perhaps it would be better to have the note like "after the -nai stem of a verb; for サ変 as せ～ず; etc. " ?
      I don't think we should go down the path of giving an exhaustive listing of the verb classes 
      

Sentence E: 懸命に頑張ったが、我々の抗議は認められ（ず）、得点も入らなかった。
   "Though we put all our might into it, our protests weren't noticed [, nor did we hit our goals/and we didn't hit our goals either]"

What about ず as a coordinating conjunction indicating consequence? Sentence E could also be interpreted this way, could it not?
"Though we put all our might into it, our protests weren't noticed, so we also didn't hit our goals."

Or here as a conjunction indicating reason.
Sentence F: 新しい政府がどのような方針で対処するかは予想もつかず、 不安な気持に陥ることも多い。
Since one can hardly anticipate by which principles a new government will act, it often leads to uneasiness.


I think I was also a bit biased towards "without doing" because, if I recall correctly, that was the situation in which I first encountered conjunctive ず.
I see now that ずに is limited to "without doing", but for plain ず that's just one possible interpretation. 
The general dominance of ず throughout all examples in the paper in itself speaks for the versatility.
Maybe it would be helpful to list a few different options as glosses, with the most "neutral" one first?

"not... and..."
"not doing..."
"not... so..."
("without doing")

I too find it difficult to gloss this, e.g. in these sentences there is hardly a gloss to give:
「連絡が取れず、心配した」Unable to make contact, I was worried.

「何も言わず、動きもしない」Not saying a word, nor moving an inch.

@@ -13,0 +14,3 @@
+<gloss>(does) not... and...</gloss>
+<gloss>not doing</gloss>
+<gloss>not... so...</gloss>

I think repetition in the notes detracts from the overall quality of the entry. Also, shouldn’t the archaic form be noted as connecting to archaic nidan verbs as well (e.g., 捨つ->捨てず)?

It brings a tear to the eye... 

Anyhow, there are another 2 paragraphs of argumentation under the rejected version of this entry not-so-briefly raising some points as to why this entry was a problem besides the one Sombrero just brought up - including the reason why I am now editing this recently approved entry (so sorry about that).
In short, it is the サ変 -nai stem problem again - I once again would like to address it, if nothing else.
The note might seem overly long but ironically is shorter than other alternatives I have tried as the kanji save us a handful of characters.

The above is why I edited the entry, now onto the reason why this is 2 paragraphs long...

ず is not properly represented by "without doing" as it is not quite identical to ずに - just as なくて and なく are not quite the same as ないで.
I will try to use only the paper and two sentences Sombrero has provided to make my point. 
However, of the paper, I will only be focusing on sentences c, e, f and g of diagram 1 on page 3 as those are the ones that deal with the bare negative auxiliaries without a verb pre-attached, and they are also the only ones wherein ず、ずに、なくて、なく and ないで are all present as options.

So, diagram 1. C, e, f and especially g.
Across these ず is chosen a total of 1840 times, 436(~23.69%) in c, 539(32.5%) in e, 531(~28.85%) in f and 275(~14.94%) in g.
ずに is chosen 233 times, 221(~94.84%) in g, and 12(<6%) times across all other options.
Notice how g concentrates almost all instances of ずに and is also the one in which ず sees the least amount of usage, as well as the sharpest decrease compared to the next value above it (c).
なく and なくて were chosen only 55 and 46 times respectively, but they were *never* chosen for g.
Meanwhile, ないで was *only* chosen for g.

Proportionally, なく and なくて have a distribution more similar to　ず, while ずに's is much more similar to ないで. 

So, where am I going with this?

The idea is that ず is more like your typical verbal coordinator that is sometimes reinterpreted as subordinating if the context supports it (like the 連用形, て or our English 'and') while ずに is almost exclusively used as a subordinator for the following clause - this is why ず still had some presence in g while ずに had almost none outside of it (and why ないで had actually none).
That is also why I had reluctantly suggested "not... and..." as the gloss and left an explanatory gloss below as I myself found the translation alone too awkward, but even so did not suggest copying the ずに entry - "without" is a subordinator and ず is not necessarily.

To close this off, the sentences.
「食べずに寝た」 cannot be made into 「寝ずに食べた」 without flipping the meaning of the sentence on its head, as it is subordinating: "I slept without eating" versus "I ate without sleeping".
Meanwhile, you can freely flip 「雨もふらず風もふかぬ」 into 「風もふかず雨もふらぬ」 and the meaning is unchanged: "Rain does not fall and wind does blow" versus "Wind does not blow and rain does not fall" 
If you think about it, I think you will agree that in the second sentence you usually do not see ずに replacing ず.

I am unsure as to whether all this rambling is actually a problem, or if the context will carry the reader to the right interpretation even in cases wherein the translation seems odd at first. 
All in all, I would like to change the gloss also, but am ultimately refraining from doing so. Perhaps someone else might have thoughts on this?

Also, [conj] tag to [1].

@@ -8,0 +9 @@
+<pos>&conj;</pos>
@@ -12 +13 @@
-<s_inf>after the -nai stem of a verb; continuative form of ぬ acting as a negative conjunction</s_inf>
+<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず; continuative form of ぬ acting as a negative conjunction</s_inf>
@@ -19 +20 @@
-<s_inf>after the -nai stem of a verb</s_inf>
+<s_inf>connects to -nai stem of 五段, 一段 and カ変 verbs and to サ変 as せ～ず</s_inf>

I still think the original brief form is sufficient, but I can live with this, I guess.

＊ Sankoku:
１：
   口語の否定の助動詞「ぬ」の連用形。
    「何も買わ━、帰った」
２：｟助動(特殊型)｠
  〔文〕〔否定の助動詞〕…ない。
    「山高きが故(ゆえ)に貴(たっと)から━」

＊ GendaiShinkoku:
１：｟助動詞・特殊型｠
   打消しの意味を表す助動詞「ぬ」の連用形。…ないで
２：［古語］
   打ち消しの意味を表す。…ない

＊ Meikyo:
〚打ち消しの助動詞「ぬ」の連用形〛
１：打ち消しの意を伴って、文を中止したり、副詞的修飾をしたりする。…ないで。
「連絡が取れ━、心配した」
「何も言わ━、動きもしない」
２： 打ち消した状態であるものやさまなどを表す。
「開(あ)か━の間(ま)」
「わから━屋」
「行方知れ━」
「怖いもの知ら━」

The references above perform a sense specific split in terms of which negation system they refer to
In sense one the paradigm in question is modern (formal or colloquial) ぬ as in （〇｜ず｜ぬ・ん｜ぬ・ん｜ね｜〇），hence them noting that it's the 連用形 of ぬ.
While in sense two the paradigm in question is archaic (or formal) ず as in（（ず）｜ず｜ず｜ぬ｜ね｜〇）(which itself represents a merger of two separate older systems).
This also leads to ず use outside of that conjunctive sense being mostly regarded as archaic (gendaiSK, daijr, shinsen, oukoku), with other refs like meikyo tagging 文語 which essentially means the same.


On ず中止法：
＊ https://detail.chiebukuro.yahoo.co.jp/qa/question_detail/q12179271617

＊ https://ja.hinative.com/questions/22362238

＊ daijisen: [...] なお、現代では、連用形「ず」は中止法として主に書き言葉で用いられ、終止形は「べからず」の形で禁止の意を表すのに用いられる。

＊ nikkoku: 「ず」は、終止形のほか、連用修飾法、中止法、また、助詞「て」「は」助動詞「き」「けむ」「けり」等につづく用法があり、連用形、終止形の二形と認められる。[...]

＊ Samuel E. Martin "A reference grammar of Japanese" p. 377:
"[...] and Tabezu ni neta 'I slept without eating = I went to bed hungry'. 
The ni is optionally omittable (cf. §9. 1 . 1 2, § 14.6); that is unusual for a precopular or an adjectival noun (though yoo and mitai are similar)
except for those that are also adverbs. In this use, the V-(a)zu forms are more often written than spoken;
speakers prefer the more relaxed V-(a)nai de or V-(a)nakute to carry the same meaning, that of a negative gerund. 

But in Kansai speech V-(a)zu is still used for V-(a)naku [te] =
V-(a)nai de as in: Ame mo hurazu kaze mo hukan[u] 'It does not rain, nor does the wind
blow' (= Ame mo huranai de kaze mo hukanai) ; Ame to kaze ga hidokute, kawara ni wa
dare mo izu, kawa wa are ni areta 'The rain and wind were so terrible there was no one
along the riverbed, and the river raged and raged' (SA 2663.48a). 

For such dialects, we will say that V-(a)zu is an alternant way of making the negative infinitive, the other way
being V-(a)naku, in these dialects usually rendered as V-(a)na'u = V-(a)noo. (And the use
of the infinitive for the gerund is more general; see p. 395.) 

In standard Japanese, too, the
V-(a)zu [nil form will sometimes be preferred to the more colloquial negative gerunds,
especially in stereotyped adverbial expressions such as osimazu [ni (kane o tukau)
'(spends money) unstintingly, generously' "

＊ https://ynu.repo.nii.ac.jp/record/8096/files/35-13.pdf
  A 1996 study investigating the emergence of なく・なくて・ないで forms to replace ず and ずに conducted on university students (only pages 2-4 are really relevant to this entry).
  This isn't supposed to be a main reference but more of a means of substantiating the first sense


.

I think nonさん was on the right track. 
While 中止法 in itself is nothing too out of the ordinary, it may be useful to have a separate sense indicating this here.
Furthermore, having precisely this sense is useful to the average user of this dictionary, because it's a more likely case for this auxiliary to be encountered.

Also I don't see how this has [suf], ず isn't a derivational morpheme, it's inflectional.
There also aren't any refs to substantiate this AFAICS

Perhaps sense one could have a [conj-prt] tag, but that may also be a bit dubious at the same time.

@@ -9 +8,0 @@
-<pos>&suf;</pos>
@@ -10,0 +10,9 @@
+<xref type="see" seq="2576210">ずに</xref>
+<misc>&form;</misc>
+<s_inf>after the -nai stem of a verb; continuative form of ぬ acting as a negative conjunction</s_inf>
+<gloss>without doing</gloss>
+</sense>
+<sense>
+<pos>&aux-v;</pos>
+<xref type="see" seq="2257550">ない</xref>
+<misc>&arch;</misc>
@@ -12 +20 @@
-<gloss>not doing</gloss>
+<gloss>not</gloss>

I don't think the proposed revision is very useful to the typical user of this dictionary. There may be scope somewhere for detailed etymological and morphological information about these sorts of terms, but I don't think this is the place.
I'm reverting the entry to the 2016 original so that the comments remain visible. I'll reopen it for a while in case there is further discussion.

@@ -10 +9,0 @@
-<pos>&conj;</pos>
@@ -14,11 +13,2 @@
-<s_inf>連用形 of ～ぬ acting as a conjunction</s_inf>
-<gloss g_type="expl">used to negate a verb and continue onto next sentence</gloss>
-</sense>
-<sense>
-<pos>&aux-v;</pos>
-<xref type="see" seq="2441300">ぬ・1</xref>
-<xref type="see" seq="2441300">ぬ</xref>
-<xref type="see" seq="2441300">ぬ</xref>
-<misc>&form;</misc>
-<s_inf>used at sentence end; old 終止形 of ～ぬ</s_inf>
-<gloss>not</gloss>
+<s_inf>after the -nai stem of a verb</s_inf>
+<gloss>not doing</gloss>

After some consideration, the translation on [1] should be changed as conjunctive ず is not properly represented by it: this ず conflates coordination and negation into a single morpheme, english has no such morpheme and thus lacks a direct translation. I have considered something such as "does/will not... and..." but that seemed confusing without an example sentence so I opted for an explanatory definition that is, hopefully, clear enough.

I realise I have forgotten to say why I have removed the notes on connection: I did so as I assume the notes will compel a reader to click the cross-reference to ぬ, leading them to learn that from there.

An additional idea that came to mind: this entry could be merged with ぬ by adding ず as a reading and adding this glossary to its, it would only need a few "as ～ず" or "as ～ぬ" to be added to the notes, the same could be done with ん. This practice could be applied to merge other entries that are variations of the same thing and with that trim down the number of individual entries with the benefit of consolidating currently scattered items under their corresponding set.

@@ -15 +15 @@
-<gloss>not</gloss>
+<gloss g_type="expl">used to negate a verb and continue onto next sentence</gloss>

The notes only do two things: tell you what it is and tell you what it does. Do you presuppose that a reader does not wish to know either or neither of these things?
Of course shoving 連用形 or "conjunction" in there will not automatically allow them to directly download information concerning those concepts unto their brain, I truly hope you did not think this to have been my intent. 
The terms are put there so that the reader may research them on their own, they can even be researched back with this dictionary if someone wishes to research through translation. Could you construct an argument against these points, so that we can have a productive discussion and make a better entry?
The information is not there to "cater to grammar nerds" as you so creatively put, it is there to teach the language: grammar plays an important part of second-language acquisition for many, especially adults, even more so when there is such a disparity between the structures of Japanese and several Indo-European languages. Did you learn Japanese without studying any form of grammar? I would doubt that unless you grew up using Japanese.

"連用形" etc. isn't helpful. 99% of users won't know what it's a reference to. If you really were as smart as you seem to think, you'd realize the goal here isn't to cater to the 1% elite grammer nerds

daijs, daijr

Entry was missing the information that ず can form a conjunction.
Added second sense as terminal ず can still be found in set phrases as well as certain media.
Opted for 連用形 and 終止形 over translations as the original Japanese can likely be researched with less potential of wrong results than something such as connective form or terminal form, they can also be looked up within the dictionary whereas translations cannot.

@@ -9,0 +10 @@
+<pos>&conj;</pos>
@@ -13,2 +14,11 @@
-<s_inf>after the -nai stem of a verb</s_inf>
-<gloss>not doing</gloss>
+<s_inf>連用形 of ～ぬ acting as a conjunction</s_inf>
+<gloss>not</gloss>
+</sense>
+<sense>
+<pos>&aux-v;</pos>
+<xref type="see" seq="2441300">ぬ・1</xref>
+<xref type="see" seq="2441300">ぬ</xref>
+<xref type="see" seq="2441300">ぬ</xref>
+<misc>&form;</misc>
+<s_inf>used at sentence end; old 終止形 of ～ぬ</s_inf>
+<gloss>not</gloss>

koj, etc.