Rob Browning [Sun, 26 Mar 2023 18:06:49 +0000 (13:06 -0500)]
Generate README.Debian from patch README-Debian psuedo-headers
Rework debian/patch-to-news to generate the README.Debian entry from a
README-Debian DEP-3 style psuedo-header, when present. The header's
syntax is similar to the debian/control Description format. Among
other things, this change makes the common case of cherry-picking
upstream patches easier, and less invasive, since in many cases, a
simple README-Debian header can be appended while adding the other
headers, instead of (as previously) reindenting the upstream commit,
etc.
Rob Browning [Sun, 31 Dec 2023 07:24:38 +0000 (01:24 -0600)]
Give each emacs flavor its own emacsclient
Move emacsclient from emacs-bin-common to the emacsVER (i.e.
emacs-lucid, emacs-gtk, ...) packages so that each one will have the
binary that was built for that flavor.
I was using emacs-lucid, and after upgrading to 29.1,
emacsclient (which was the pgtk flavor) no longer worked correctly,
for example "-c -n" didn't open new frames. However the lucid build
emacsclient worked fine.
Explicitly list the variant and common bin alternatives instead of
globbing. Add the alternatives variables to deb_sub so we don't need
the separate perl -i substitutions.
Stop splitting some of the debian/rules $(call ...) invocations across
two lines because it was introducing a space into the argument value
which would then break quoted expansions like "$(1)...".
Sean Whitton [Tue, 15 Nov 2022 23:26:59 +0000 (16:26 -0700)]
Mark flaky test process-tests/multiple-threads-waiting as unstable
It times out:
passed 22/28 process-tests/fd-setsize-no-crash/make-serial-process (0.021449 sec)
make[5]: *** [Makefile:182: src/process-tests.log] Error 134
GEN src/regex-emacs-tests.log
GEN src/search-tests.log
GEN src/syntax-tests.log
GEN src/textprop-tests.log
GEN src/thread-tests.log
GEN src/timefns-tests.log
GEN src/undo-tests.log
GEN src/xdisp-tests.log
GEN src/xfaces-tests.log
GEN src/xml-tests.log
make[5]: Leaving directory '/<<PKGBUILDDIR>>/debian/build-gtk/test'
make[4]: [Makefile:335: check-doit] Error 2 (ignored)
SUMMARY OF TEST RESULTS
-----------------------
Files examined: 375
Ran 5408 tests, 28 failed to run, 5273 results as expected, 0 unexpected, 135 skipped
1 files did not finish:
src/process-tests.log
make[4]: *** [Makefile:336: check-doit] Error 2
Rob Browning [Sun, 16 Dec 2018 23:13:36 +0000 (17:13 -0600)]
Mark vc-bzr-test-fauilt-bzr-autoloads as unstable for now
Currently the test fails like this:
Running 3 tests (2018-12-17 12:17:43-0600)
passed 1/3 vc-bzr-test-bug9726
Mark set
Press C-c C-c when you are done editing.
Enter a change comment. Type C-c C-c when done
passed 2/3 vc-bzr-test-bug9781
Falling back on "slow" status detection ((file-missing "Opening input file" "No such file or directory" "/tmp/vc-bzr-testVlgmsb/bzr/.bzr/checkout/dirstate"))
Error: (error "Running bzr status --no-classify loaddefs.el...FAILED (status 3)")
Warnings in `bzr' output: bzr: ERROR: invalid header line: ''
Error: (error "Running bzr status --no-classify loaddefs.el...FAILED (status 3)")
Warnings in `bzr' output: bzr: ERROR: invalid header line: ''
Rob Browning [Mon, 10 Oct 2016 22:35:56 +0000 (17:35 -0500)]
Kill gpg agent in package-test.el to avoid a race
package-test.el should no longer fail during clean up.
* Previously a delete-directories call raced with the gpg agent's own
cleanup process (presumably triggered by the first deletion of one of
the agent's sockets). As a result, it looks like the agent might
delete one of its sockets after delete-directories had decided to
delete the socket, but before it made the attempt, causing an
exception.
* To fix the problem, explicitly ask gpg-connect-agent to kill the agent
before attempting to delete the gnupg home directory, and then delete
via "rm -rf" to ignore any vanishing files.
Rob Browning [Tue, 5 Apr 2011 03:46:24 +0000 (22:46 -0500)]
Remove files that appear to be incompatible with the DFSG
Files that appear to be incompatible with the DFSG have been removed.
A number of files have been removed from this package because their
licenses are not compatible with the Debian Free Software Guidelines
(DFSG), or because it wasn't completely clear that their licenses are
compatible.
In particular, all of the files which are covered under the GFDL and
have invariant sections have been removed in accordance with this
General Resolution: http://www.debian.org/vote/2006/vote_001.
The files that have been removed, but still appear to be
distributable, have been moved to packages in Debian's non-free
section.
Author: Rob Browning <rlb@defaultvalue.org> Added-by: Rob Browning <rlb@defaultvalue.org>
Status: new
Update CMake support due to upstream changes (bug#64922)
A recent change in tree-sitter-cmake grammar support for CMake (commit fe9b5e0), now put arguments are wrapped in a new argument_list node.
To support the old and new version of the grammar, a new function was
added on which string syntax highlighting now depends.
* lisp/progmodes/cmake-ts-mode.el
(cmake-ts-mode--font-lock-compatibility-fe9b5e0): Indent helper
function to handle different tree-sitter-cmake version.
* lisp/progmodes/cmake-ts-mode.el
(cmake-ts-mode--font-lock-settings): Use the new function to handle
the new argument_list node.
Remove nullptr named node from c++-ts-mode (bug#64818)
The nullptr node was changed from a named node to an unnamed node
upstream[0], which caused font locking to break. As this is a small
enough regression, no compat code is required.
* lisp/progmodes/c-ts-mode.el (c-ts-mode--font-lock-settings): Remove
node no longer in use.
* lisp/progmodes/typescript-ts-mode.el
(tsx-ts-mode--font-lock-compatibility-bb1f97b):
Add argument so that we run the 'treesit-query-capture' when the
language is 'typescript', not only 'tsx'.
* lisp/progmodes/typescript-ts-mode.el
(typescript-ts-mode--font-lock-settings): Use supplied argument.
Update JSX support due to upstream changes (bug#64647)
A recent change in tree-sitter-javascript grammar support for
JSX (commit bb1f97b), changed two things:
1. renamed nested_identifier to member_expression
2. removed jsx_fragment, jsx_text is used instead
* lisp/progmodes/js.el (js-jsx--treesit-indent-compatibility-bb1f97b):
Indent helper function to handle different tree-sitter-javascript
version.
* lisp/progmodes/js.el (js--treesit-indent-rules): Use the new
function to handle both jsx_fragment and jsx_text.
* lisp/progmodes/js.el (js-jsx--treesit-font-lock-compatibility-bb1f97b):
Font lock helper function for handle different tree-sitter-javascript
version.
* lisp/progmodes/js.el (js--treesit-font-lock-settings): Use the new
function to handle both nested_identifier and member_expression.
Update TSX support due to upstream changes (bug#64647)
A recent change in tree-sitter-typescript grammar support for
TSX (commit b893426), changed two things:
1. renamed nested_identifier to member_expression
2. removed jsx_fragment, jsx_text is used instead
* lisp/progmodes/typescript-ts-mode.el
(tsx-ts-mode--indent-compatibility-b893426): Indent helper function
to handle different tree-sitter-typescript version.
* lisp/progmodes/typescript-ts-mode.el
(typescript-ts-mode--indent-rules): use the new function to handle
both jsx_fragment and jsx_text.
* lisp/progmodes/typescript-ts-mode.el
(tsx-ts-mode--font-lock-compatibility-bb1f97b): Font lock helper
function for handle different tree-sitter-typescript version.
* lisp/progmodes/typescript-ts-mode.el
(typescript-ts-mode--font-lock-settings): Use the new function to
handle both nested_identifier and member_expression.
Eli Zaretskii [Sat, 22 Jul 2023 08:02:04 +0000 (11:02 +0300)]
; Improve documentation of automatic addition to minibuffer history
* doc/lispref/minibuf.texi (Minibuffer History)
(Minibuffer Commands, Text from Minibuffer)
(Minibuffer Completion): Document the behavior of 'M-n' and
"future history" when the value(s) in DEFAULT is/are exhausted.
(Bug#64656)
Ignore quit while getting interprogram paste in kill-new
On X, if the current selection owner is not responding to selection
requests, the user may want to take ownership of the selection. The
obvious way to do this is to kill some text (which a user might also
be doing just as part of normal editing at the time the selection
owner becomes nonresponsive). However, if
save-interprogram-paste-before-kill is non-nil, then killing text will
hang until the user quits, and this quit will abort the entire
kill-new, preventing the user from taking ownership of the selection.
Now instead if the user quits while we are attempting to retrieve the
selection from hanging owner, we will proceed to take ownership of the
selection as normal, resolving the problem.
(One example of a selction owner that might not be responding to
selection requests is another instance of Emacs itself; while Emacs is
blocked in call-process or Lisp execution, it currently does not
respond to selection requests.)
* lisp/simple.el (kill-new): Ignore quit while getting interprogram
paste (bug#64423)
* configure.ac (netbsd): Don't set TERMINFO=no unless the termcap
library is either -ltermcap or -lcurses. This prevents aborts
because on recent versions of NetBSD libtermcap is actually a
symlink to libterminfo. (Bug#64577)
projects / emacs.git / log