Some proposed backports from master to emacs-25

Some proposed backports from master to emacs-25

[npostavs]

[-- Attachment #1: Type: text/plain, Size: 377 bytes --]


I had marked some bugs as fixed in 25.2 and pushed to master because I
didn't know at the time it was going to be 26.  I think there were some
others working under the same assumption, so I've gathered those fixes
to backport in a semi-automated fashion by searching commit messages in
the range emacs-25..master for non-wishlist bugs marked fixed in 25.2,
here is the code:


[-- Attachment #2: bug gathering code --]
[-- Type: text/plain, Size: 1896 bytes --]

(require 'debbugs)

(defvar backporting-fixed-bug-list
  (debbugs-get-bugs :status "done" :tag "fixed"
                    :archive "both"))

(defvar backporting-fixed-bug-status-list
  (apply #'debbugs-get-status backporting-fixed-bug-list))

(defun backporting-fixed-25.2-bug-list ()
    (cl-loop for bug in backporting-fixed-bug-status-list
             if (and (member "25.2" (cdr (assq 'fixed_versions bug)))
                     (not (equal "wishlist" (cdr (assq 'severity bug)))))
             collect (cdr (assq 'id bug))))

(mapconcat #'number-to-string (backporting-fixed-25.2-bug-list) "\\|")

;;; Actually I think this misses a call to tac, because it seemed that
;;; the commits were picked in reverse order (I did a rebase to put
;;; the in right order afterwards).
;; git log --format='%H' --grep='16294\|16345\|16390\|16406\|16483\|16513\|16904\|17039\|17582\|17707\|17716\|17738\|17989\|17999\|18024\|18028\|18089\|18092\|18110\|18202\|18211\|18279\|18527\|18634\|18809\|18829\|19114\|19152\|19209\|19213\|19214\|19215\|19255\|19368\|19497\|19587\|19638\|19722\|19754\|19801\|19851\|20038\|20158\|20181\|20304\|20460\|20520\|20654\|20702\|20724\|21002\|21024\|21091\|21171\|21225\|21231\|21252\|21269\|21552\|21576\|21577\|21601\|21706\|21851\|21852\|21881\|21936\|21962\|22117\|22140\|22170\|22325\|22530\|22531\|22576\|22583\|22586\|22592\|22594\|22596\|22648\|22664\|22724\|22764\|22824\|22827\|22837\|22841\|22928\|22964\|23020\|23071\|23116\|23167\|23262\|23290\|23374\|23390\|23401\|23411\|23459\|23949\|23955\|24133\|24166\|24257\|24308\|24315\|10540\|10723\|10942\|11400\|12378\|12939\|13187\|13269\|13571\|13745\|14256\|14554\|14577\|14687\|14854\|14919\|15021\|15047\|15171\|15909\|3137\|3393\|4589\|5001\|5262\|5314\|5479\|5650\|5727\|7522\|7751\|8634\|8693\|8925\|9342\|9730' gnu/emacs-25..gnu/master | git cherry-pick -x --allow-empty --stdin


[-- Attachment #3: Type: text/plain, Size: 739 bytes --]


This turned up 129 commits, 54 of which were already applied (i.e., the
resulting diff from cherry-picking was empty), and 26 that produced
conflicts, 2 that looked like a new feature (and 1 which is a merge
commit).  I've pushed the remaining 46 to scratch/backports-25.2.

Git added "cherry picked from commit" but it looks like
gitmerge-skip-regexp wouldn't match that, should I change it to match?

I've the listed the skipped commits below.  Apart from 'Revert "Cleanup
tooltips"', which I think just got listed because I used a fairly lax
regexp (didn't want to miss commits using formats other than Bug#xxxxx),
it could make sense to backport the ones 'Skipped due to conflicts' too
if someone feels like resolving the conflicts.


[-- Attachment #4: log of dropped commits --]
[-- Type: text/plain, Size: 5793 bytes --]

# Skipped, looks like new feature
a3b4f6e Add a variable to control "shorter command" suggestions
36eff6b Document extended-command-suggest-shorter

# Skipped due to conflicts
54b7eb0 Revert "Cleanup tooltips"
73071b4 Support srcset in <img>
247c388 Describe WHEN in all the define-obsolete- macros
970b82b Minor doc fix for ImageMagick/SVG builds
207a314 Fill the completion-table-with-predicate doc string
9392193 Transform mentions of `eval-after-load' to `with-eval-after-load'
e4c2627 Change all occurrences of "Mouse-[0-9]" to "mouse-[0-9]"
65797b1 Make icomplete respect `completion-ignored-extensions'
ad3ef41 Clarify some doc strings
7b9a5c2 Fix: (void-variable url-http-response-status)
f069d85 ; Run find-function-after-hook after finding a symbol
ff9de7b Pass in port only if provided by caller in the OS X keychain
6e70769 Add a Catalan language environment
1ba0322 Make the describe-key prompts clearer
cede502 Allow scrolling while querying multiple choice
46ef01f Fix encoding problem introduced by previous patch series
d0c2957 Fix typo in Gnus regexp
ebc6985 Use an X popup in read-multiple-choice if running from a mouse command
fa6912f Display cursor in echo area when prompting
e49c46f Make NSM prompting clearer
d686802 Reconnect erc even on server errors
1fe7344 Disconnection fixes for erc
d50b89d Further erc asynch fixes
5917b50 Use a different port for TLS erc
2a0f18d Always save eww history
cdaf330 Allow several <tbody> tags in shr

# Already applied (the diff was empty)
75808adbc185a0cb13993cfb25f61fbd31b074a9 Fix debugging of string-match-p errors
4cf75ccb20d7a5cf74d9af5759c3aa727efe7b99 cua-prefix-override-inhibit-delay doc fix
e2eca5098bf0677dbd1ede84f58159a95859d195 global-eldoc-mode doc fix
a11cff1f956e23b06c58c0495ab97f7770e560dc normal-top-level-add-subdirs-to-load-path doc fix
c2dfac3dd653a05ae51a12cf4402aafa29ab1771 isearch-search-fun-function doc tweak
a36522bad55ec216a490a42702926626d5577da6 Fill some imenu--index-alist doc lines
24e10fa467c42ca7642e578ec33119488db5d533 Fmarker_position doc string clarification
7f05c020004f01eabd104d8a6547544c62fa0398 Mention with-silent-modifications in the lispref manual
983c98391763698834234e98d2c51880660ff94c with-silent-modifications doc clarification
9d3a12c3926346933c0213e4c83c388eedaa79af Add a cross ref to Optional Mode Line
205a0abccf9554a2461ad8f80ec8caa9ac25c714 Add a doc string to display-time-string
41518302bed10884014251bcd262c4006ccb42b7 custom-buffer-style doc fix
8b1ee00a56edc9a7dae3b6df0485a2dbe8d65065 Document how to check for ImageMagick support
62fa56189e00bed449929e667ef64bffa375d691 Rearrange the doc of query-replace slightly
aa49452fdcb5bad6d894c89ef541ff4a3ebbdeb0 apropos-print doc fix
3515004631e00fe42a7cc8749ac27bff0c8c67b3 completion-table-with-predicate doc string fix
46437046e4a07bdf34f363198f7ca019051ecefe replace-match-maybe-edit doc clarification
b50ddb60938371728e7969dd9cfccc167a1f0355 add-timeout doc fix
46bf403dd58b52e575f6f095a19d657307d61139 Extremely minor doc fix in Choosing Window
674e2804039e576254b2b7fee44eb0adef29fc3b cursor-type doc fix
2fe4e5a8133b21e08adb1632c61651673a3034a1 Add a link from Tool Bar to Images
dd804d5808b348b86dfa845e74d74c8b269f01bf Fcompare_buffer_substrings doc string clarification
72d1393c751c6b6ef0841b7c308e8c52a6d5f605 Fix custom types for cursor-in-non-selected-windows
e49942c6358298253630ef5e80edb1d9a594c121 Doc clarification to mwheel-scroll
e45056d19dcb87d86274c33681e21e145fee9e9a Clarify the doc of eval-expression-print-format
5fcf1deb453911b57395f58bf33b9d29e0524087 Explictly explain that package-initialize loads the packages
6fb6f201aab3f17022a67f15c9623462d407b301 Have the doc strings of `load-path' and `require' mention each other
ed8d298c52a77971420e62c1e0892ae4c6fdc0c1 Doc fix for insert-pair-alist
4f199d47f139001261cfdcff54da30d036b8081c Doc fix for font-lock-remove-keywords
505b0a3ee3374b8c9b268315bde056b7bfa81f0a Doc fixed for next-error-buffer-p
d72f4e07917b955503349b686a2330727b61eccb Doc fix
2e71b7a0eb098cba200c9eab86d552fb7812dc33 delsel doc touch ups
67007e282ec08cb4d143114e361cd6b403c40ea2 Doc fixes for menu-bar.el
d1aab04ec3e5d6f4cc6eee4d248acdff114b364c Minor doc clarification
0baf1a2ec2cd1d39f20345e064422821e6837e5e Wrap the auto-generated doc string
1a3f1d06f51dbc187fec99cde3fd3743e4dd07f8 Clarify hi-lock-find-patterns
8b7fb43e375f34afaef5cb1770ecdc5848036d4d Add a doc string to `winner-mode'
83871e2d16b67f13d18df34795eba1dde2c522a4 Doc fix for align-newline-and-indent
ca06d2a7140b2cc4ca2f89ccc24a8003d1c4cd88 Doc fix for `kbd'
7b3738ebb5a6ec9c5f333fe5767926cbceac0b8a Doc tweak
b8c50b27a26d58b327a8e49f01d478f1c39c1213 find-lisp doc touchups
df2910e61d8a7870efc476411504d481d5b3ca06 Don't have the manual claim that it lists all CL incompatibilities
5196057f2060bc3853c2e81d0187472307000b69 Clarify the `interactive' doc string slightly
ca68c6ed7b10e0de427ada355c7b8be225677838 Mention `lisp-indent-function' in the lispref manual
86f03209d7188d2a3fd5e3ae1736102bcc7b7265 Clarify doc string of internal compilation function
a39e6264c084b208f61fdd7d19d5056e37eaee7f Ignore invalid base64 encoded embedded images
a213bb6f7936e77c4b8b10b6ebac702f7802fc31 Render empty <ul><li><ul> correctly
4c6197db87550d776d52242b555d85e9ae2521d4 Declare $ as an expression prefix in SCSS
4e636973943092916164edf27f9b133279f6ab0c Use memmove instead of memcpy on overlapping regions
d353b9d9524da23275e187577de453b64572d525 Make buttons in header lines work
04c32cce99298df4b9d8bed6d1baa9926ab938d6 Remove spurious newline in package list
83fdcd23e79f7b4f4f55c68400e5463d1146d87a Make toggling checkboxes work again
27d0c51d3f024843d6cfd4fbc79fb155e7fe304f Don't store cookies with empty names
0e9e8ca5a33f424f8fe5faab5e2f00fa9643644f Stop rendering HTML before specdlr exhaustion

Re: Some proposed backports from master to emacs-25

[Eli Zaretskii]

> From: npostavs@users.sourceforge.net
> Date: Sun, 02 Oct 2016 10:05:50 -0400
> 
> I had marked some bugs as fixed in 25.2 and pushed to master because I
> didn't know at the time it was going to be 26.  I think there were some
> others working under the same assumption, so I've gathered those fixes
> to backport in a semi-automated fashion by searching commit messages in
> the range emacs-25..master for non-wishlist bugs marked fixed in 25.2,

Could you please provide a short explanation why you think each of the
backports should be in 25.2?  I've looked through the log messages,
and couldn't figure that out easily.

In general, since we want 25.2 released quickly, I think the criteria
for back-porting should be somewhat rigid, something like:

  . any doc fixes
  . regressions from 24.5
  . problems introduced in features new in 25.1

> (defvar backporting-fixed-bug-list
>   (debbugs-get-bugs :status "done" :tag "fixed"
>                     :archive "both"))
> 
> (defvar backporting-fixed-bug-status-list
>   (apply #'debbugs-get-status backporting-fixed-bug-list))
> 
> (defun backporting-fixed-25.2-bug-list ()
>     (cl-loop for bug in backporting-fixed-bug-status-list
>              if (and (member "25.2" (cdr (assq 'fixed_versions bug)))
>                      (not (equal "wishlist" (cdr (assq 'severity bug)))))
>              collect (cdr (assq 'id bug))))

That looks quite a bit more than the above criteria, but maybe I'm
missing something.

Thanks.

Re: Some proposed backports from master to emacs-25

[Noam Postavsky]

[-- Attachment #1: Type: text/plain, Size: 800 bytes --]

On Sun, Oct 2, 2016 at 10:54 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> Could you please provide a short explanation why you think each of the
> backports should be in 25.2?  I've looked through the log messages,
> and couldn't figure that out easily.
>
> In general, since we want 25.2 released quickly, I think the criteria
> for back-porting should be somewhat rigid, something like:
>
>   . any doc fixes
>   . regressions from 24.5
>   . problems introduced in features new in 25.1

Ah, in that case, I would propose just these 4 (pushed to
scratch/backports-25.2-strict)

;;; Regression in Emacs 25.1
Don't require isearch-update before isearch-done

;;; This bug actually crashes in emacs-25
Avoid error in icalendar--read-element

;;; Doc fixes
Make a menu less ambiguous
Doc fix for url-http

[-- Attachment #2: backporting-strict.log --]
[-- Type: text/plain, Size: 2637 bytes --]

;;; Regression in Emacs 25.1
eb5dd2c Don't require isearch-update before isearch-done

;;; This bug actually crashes in emacs-25
171ec16 Avoid error in icalendar--read-element

;;; Doc fixes
f3a19cf Make a menu less ambiguous
25895c4 Doc fix for url-http

;;; Drop these
e1e768e Default web pages to right-to-left
a33eb50 Let url use default file modes when copying files
7d89389 Make erc connect asynchronously
145e369 Make the nsm query say what it did after the user interaction
8fd0223 Make mail-extract-address-components return the user name more
aad7ede Ensure progress when fetching from the queue
2284444 Require that the buffer that we're trying to patch exist
b16e6c3 Require that the buffer that we're trying to patch exist
5c084d9 Remove spurious newline in package list
fbe5f7b Completely ignore buffers with no name when copying file names
95d5520 Allow optional parameter to be nil
5266c11 Allow more shell script defun forms
1cd6309 Allow ff-find-other-file (etc) to work with indirect clone buffers
70244eb Allow highlighting things like @math{2^{12}}
0780aa5 Replace XXX acronyms with draft standard ones
344bc2b Strip out some leading whitespace when looking at logs
5f4c54b Declare $ as an expression prefix in SCSS
0d9709e Meta + shift + Fn combos are recognized by xterm>=216.
e2e25ae Make <div> in <li> not insert extra newlines
80cf0b6 Restore the fix to bug#18527 from commit d6868025
30376cc Allow _ characters in SQL prompts
69e0e89 Handle auth-source items with special characters on OS X
77c3f7f Make `q' etc work in image mode again
d7de0c4 smtpmail would say it's done before it is
f24a305 Make `undo' of `C-x r t' put point where it should be
f65b1a3 Don't consider colons to be paragraphs starting chars in strings
51656da Fix number-at-point in lisp buffers
d95942f Clean up custom-buffer-create code slightly
18f1ab7 Have describe-variable output multi-line values better
7826f1f Make ibuffer not break on newline-embedded buffer names
5afea13 Rename "Directory Search" menu to "Servers"
d6f5232 Remove ": ?" from the read-face-name prompt
87a7490 Clarify `read-face-name' doc and tweak the code
3713acd Cancel the eldoc timer when switching off eldoc mode
33b69a1 Warning fix in jit-lock-mode
c9443ca Make describe-variable look up the variable in the current buffer
4af6945 Add `size-indication-mode' to the menu on `mouse-1' "Top"
90262cc Allow `global-set-key' to bind keys under the `M-o' map
12d1955 * lisp/files.el (version-control): Drop :group vc (Bug #14687)
cf54324 Improve error when installing non-package dirs
0876366 Fix back-white <-> black-white typo
9e9f7c8 Don't --load directories

Re: Some proposed backports from master to emacs-25

[Eli Zaretskii]

> From: Noam Postavsky <npostavs@users.sourceforge.net>
> Date: Sun, 2 Oct 2016 12:40:24 -0400
> Cc: Emacs developers <emacs-devel@gnu.org>
> 
> >   . any doc fixes
> >   . regressions from 24.5
> >   . problems introduced in features new in 25.1
> 
> Ah, in that case, I would propose just these 4 (pushed to
> scratch/backports-25.2-strict)
> 
> ;;; Regression in Emacs 25.1
> Don't require isearch-update before isearch-done
> 
> ;;; This bug actually crashes in emacs-25
> Avoid error in icalendar--read-element
> 
> ;;; Doc fixes
> Make a menu less ambiguous
> Doc fix for url-http

These are fine with me; please push to emacs-25.

Thanks.

Re: Some proposed backports from master to emacs-25

[Noam Postavsky]

On Sun, Oct 2, 2016 at 10:05 AM,  <npostavs@users.sourceforge.net> wrote:
>
> Git added "cherry picked from commit" but it looks like
> gitmerge-skip-regexp wouldn't match that, should I change it to match?
>

What about this question?

Re: Some proposed backports from master to emacs-25

[David Engster]

Noam Postavsky writes:
> On Sun, Oct 2, 2016 at 10:05 AM,  <npostavs@users.sourceforge.net> wrote:
>>
>> Git added "cherry picked from commit" but it looks like
>> gitmerge-skip-regexp wouldn't match that, should I change it to match?
>>
>
> What about this question?

If the cherry-pick succeeds without conflict, git can detect them
through the patch-id and add a cherry-mark, which in turn will make
gitmerge ignore those commits.

If there is a conflict, you will have to manually commit anyway and you
should add 'backport' to the commit message.

-David

Re: Some proposed backports from master to emacs-25

[Eli Zaretskii]

> From: Noam Postavsky <npostavs@users.sourceforge.net>
> Date: Mon, 3 Oct 2016 18:59:50 -0400
> 
> On Sun, Oct 2, 2016 at 10:05 AM,  <npostavs@users.sourceforge.net> wrote:
> >
> > Git added "cherry picked from commit" but it looks like
> > gitmerge-skip-regexp wouldn't match that, should I change it to match?
> >
> 
> What about this question?

Sorry, forgot to answer.  I see no problems, as gitmerge-skip-regexp
already handles that:

  (defvar gitmerge-skip-regexp
    ;; We used to include "sync" in there, but in my experience it only
    ;; caused false positives.  --Stef
    "back[- ]?port\\|cherry picked from commit\\|\\(do not\\|no need to\\) merge\\|\
                     ^^^^^^^^^^^^^^^^^^^^^^^^^
  re-?generate\\|bump version\\|from trunk\\|Auto-commit"
    "Regexp matching logs of revisions that might be skipped.
  `gitmerge-missing' will ask you if it should skip any matches.")

Re: Some proposed backports from master to emacs-25

[David Engster]

Eli Zaretskii writes:
>> From: Noam Postavsky <npostavs@users.sourceforge.net>
>> Date: Mon, 3 Oct 2016 18:59:50 -0400
>> 
>
>> On Sun, Oct 2, 2016 at 10:05 AM,  <npostavs@users.sourceforge.net> wrote:
>> >
>> > Git added "cherry picked from commit" but it looks like
>> > gitmerge-skip-regexp wouldn't match that, should I change it to match?
>> >
>> 
>> What about this question?
>
> Sorry, forgot to answer.  I see no problems, as gitmerge-skip-regexp
> already handles that:
>
>   (defvar gitmerge-skip-regexp
>     ;; We used to include "sync" in there, but in my experience it only
>     ;; caused false positives.  --Stef
>     "back[- ]?port\\|cherry picked from commit\\|\\(do not\\|no need to\\) merge\\|\
>                      ^^^^^^^^^^^^^^^^^^^^^^^^^
>   re-?generate\\|bump version\\|from trunk\\|Auto-commit"
>     "Regexp matching logs of revisions that might be skipped.
>   `gitmerge-missing' will ask you if it should skip any matches.")

This is only on the master branch, though. I've no idea why I didn't
commit this to emacs-25...

-David

Re: Some proposed backports from master to emacs-25

[Eli Zaretskii]

> From: David Engster <deng@randomsample.de>
> Cc: Noam Postavsky <npostavs@users.sourceforge.net>,  emacs-devel@gnu.org
> Date: Tue, 04 Oct 2016 08:51:51 +0200
> 
> >   (defvar gitmerge-skip-regexp
> >     ;; We used to include "sync" in there, but in my experience it only
> >     ;; caused false positives.  --Stef
> >     "back[- ]?port\\|cherry picked from commit\\|\\(do not\\|no need to\\) merge\\|\
> >                      ^^^^^^^^^^^^^^^^^^^^^^^^^
> >   re-?generate\\|bump version\\|from trunk\\|Auto-commit"
> >     "Regexp matching logs of revisions that might be skipped.
> >   `gitmerge-missing' will ask you if it should skip any matches.")
> 
> This is only on the master branch, though. I've no idea why I didn't
> commit this to emacs-25...

The merges are only done on master, right?

Re: Some proposed backports from master to emacs-25

[Noam Postavsky]

On Tue, Oct 4, 2016 at 3:54 AM, Eli Zaretskii <eliz@gnu.org> wrote:
>> From: David Engster <deng@randomsample.de>
>> Cc: Noam Postavsky <npostavs@users.sourceforge.net>,  emacs-devel@gnu.org
>> Date: Tue, 04 Oct 2016 08:51:51 +0200
>>
>> >   (defvar gitmerge-skip-regexp
>> >     ;; We used to include "sync" in there, but in my experience it only
>> >     ;; caused false positives.  --Stef
>> >     "back[- ]?port\\|cherry picked from commit\\|\\(do not\\|no need to\\) merge\\|\
>> >                      ^^^^^^^^^^^^^^^^^^^^^^^^^
>> >   re-?generate\\|bump version\\|from trunk\\|Auto-commit"
>> >     "Regexp matching logs of revisions that might be skipped.
>> >   `gitmerge-missing' will ask you if it should skip any matches.")
>>
>> This is only on the master branch, though. I've no idea why I didn't
>> commit this to emacs-25...
>
> The merges are only done on master, right?

Yeah, so it should be fine, except that it's potentially a bit
confusing (I asked the question because I only looked at the emacs-25
branch).

Re: Some proposed backports from master to emacs-25

[Eli Zaretskii]

> >> This is only on the master branch, though. I've no idea why I didn't
> >> commit this to emacs-25...
> >
> > The merges are only done on master, right?
> 
> Yeah, so it should be fine, except that it's potentially a bit
> confusing (I asked the question because I only looked at the emacs-25
> branch).

I'm okay with backporting that change to emacs-25.

Re: Some proposed backports from master to emacs-25

[Noam Postavsky]

On Tue, Oct 4, 2016 at 8:18 AM, Eli Zaretskii <eliz@gnu.org> wrote:
>> >> This is only on the master branch, though. I've no idea why I didn't
>> >> commit this to emacs-25...
>> >
>> > The merges are only done on master, right?
>>
>> Yeah, so it should be fine, except that it's potentially a bit
>> confusing (I asked the question because I only looked at the emacs-25
>> branch).
>
> I'm okay with backporting that change to emacs-25.

Merged the scratch/backports-25.2-strict branch in fb12af2, and
cherrypicked the gitmerge.el change as a6e0a677.