Problem: 'commentstring' not used when adding fold marker. (Maxim Kim)
Solution: Only use empty 'comments' middle when leader is empty. (Christian
Brabandt, closesvim/vim#3670)
539328197c
Comparing `vimCommand` from Vim's runtime/syntax/vim.vim, one can see
that "augroup" and similar commands are conspicuously missing. They are
handled specially (`vimAugroupKey`, `vimAutoCmd`).
Excluding them from the generated `vimCommand` keyword list, fixes their
highlighting.
closes#9327
Install gperf using package manager instead of building it from source.
When building/installing gperf from source, its install step requires
`texi2pdf` which randomly goes missing on Travis:
cd doc; /usr/bin/make install
make[1]: Entering directory '/home/travis/nvim-deps/build/src/gperf/doc'
cd . && rm -f gperf.aux gperf.toc gperf.cp gperf.fn gperf.ky gperf.pg gperf.tp gperf.vr gperf.log gperf.cps
cd . && texi2pdf gperf.texi
/bin/sh: 1: texi2pdf: not found
It's nice to test the "bundled" deps on Travis, but that gets enough
exercise on Windows and macOS, which are the platforms that actually
need "bundled" gperf.
Problem: execute() does not work in completion of user command. (thinca)
Solution: Switch off redir_off and restore it. (Ozaki Kiichi, closesvim/vim#2492)
2095148277
Before this commit, if user does this:
let g:node_host_prog = '~/.nvm/versions/node/v11.3.0/bin/neovim-node-host'
the "~/" is not expanded to user's home directory.
`:help g:ruby_host_prog` suggests a path with "~/" so technically we
already claimed to support this.
closes https://github.com/neovim/node-client/issues/102
PR #9304 added support for functions in clipboard providers. As part of
the PR I meant to move two checks in the provider code out of an if
statement into separate statements and adding additional checks for
g:clipboard attributes - as it turns out the code is wrong and it does
not implement additional checks while it adds two conditions that make
very little sense
type(g:clipboard['copy']) #isnot# v:t_func
what would make sense would be something along the lines of
type(g:clipboard['copy']['+']) #isnot# v:t_func
but might not be what we want either, so I'm reverting this.
Up to now g:clipboard["copy"] only supported string values invoked as
system commands.
This commit enables the use of VimL functions instead. The function
signatures are the same as in provider/clipboard.vim. A clipboard
provider is expected to store and return a list of lines (i.e. the text)
and a register type (as seen in setreg()).
cache_enabled is ignored if "copy" is provided by a VimL function.
As of CMake 3.12, check_include_files() also link the check executable
against the libraries listed in CMAKE_REQUIRED_LIBRARIES. Therefore we
should unset the CMAKE_REQUIRED_* variables after each respective use to
avoid them unnecessarily bleeding into other checks.
The order was swapped in #4150 to prefer `xsel` but there wasn't a clear
explanation. Meanwhile, `xsel` has been neglected upstream.
Let's trying preferring `xclip` again, we've had a few reports of
problems with `xsel`.
closes#7237
ref #5853
ref #7449
Problem: Expression evaluation may repeat an error message. (Jason
Franklin)
Solution: Check for the value of did_emsg when giving an error
for the :execute command.
8ff5af9544
Problem: Expression evaluation may repeat an error message. (Jason
Franklin)
Solution: Increment did_emsg and check for the value when giving an error
for the echo command.
76a6345433
Problem: It is not easy to edit a script that was sourced.
Solution: Add a count to ":scriptnames", so that ":script 40" edits the
script with script ID 40.
07dc18ffa4
From test_timers.vim:
Found errors in Test_paused():
First run:
function RunTheTest[35]..Test_paused line 20: Expected range 0 - 100, but got 123
Second run:
function RunTheTest[35]..Test_paused line 20: Expected range 0 - 100, but got 106
previously: #9220
- Timer tests are less reliable on Travis CI macOS 10.12/10.13.
ref #6829
ref e39dade80b
ref de13113dc1
ref https://github.com/neovim/neovim/pull/9095#issuecomment-429603452
> We don't guarantee that a X ms timer is triggered during Y ms sleep
> for any X<Y, though I would expect the load to be really bad for this
> to happen with X=10ms, Y=40ms.
From test_alot.vim:
Found errors in Test_lambda_with_timer():
First run:
function RunTheTest[35]..Test_lambda_with_timer line 19: Expected True but got 0
Second run:
function RunTheTest[35]..Test_lambda_with_timer line 19: Expected True but got 0
previously: #9220
- Timer tests are less reliable on Travis CI macOS 10.12/10.13.
ref #6829
ref e39dade80b
ref de13113dc1
ref https://github.com/neovim/neovim/pull/9095#issuecomment-429603452
> We don't guarantee that a X ms timer is triggered during Y ms sleep
> for any X<Y, though I would expect the load to be really bad for this
> to happen with X=10ms, Y=40ms.
- Call test_garbagecollect_now(), as Vim does.
Problem: Segfault when pattern with \z() is very slow.
Solution: Check for NULL regprog. Add "nfa_fail" to test_override() to be
able to test this. Fix that 'searchhl' resets called_emsg.
bcf9442307closes#8788
Fixes https://github.com/neovim/neovim/issues/9270
---
Background info per egmontkob:
https://github.com/neovim/neovim/issues/9270#issuecomment-441979176
For undercurl, the newly invented escape sequence is `4:3` strictly with
a colon, as with a semicolon it means single underlined and italic.
For colored underline, the newly invented escape sequence `58:...` is
meant to follow the pattern of `38` and `48`. [ITU
T.416](https://www.itu.int/rec/T-REC-T.416-199303-I/en) § 13.1.8 clearly
specifies the colon only as the separator (and the well-known ECMA-48
§ 8.3.117 just points to this standard).
Using semicolon instead was/is a frequent misinterpretation of this
standard, and is commonly used in the wild – for 38 and 48. More and
more emulators are catching up and beginning to support colon, in
addition to semicolon. Semicolon is pretty fragile; in case an emulator
doesn't recognize a sequence (let's say doesn't recognize the new
extension of `58`), subsequent numbers are interpreted as other
attributes. E.g. if 256-color mode is chosen then the next numeric
parameter is `5` which turns on blinking.
So, luckily, the standard is the technically better solution, the
frequent practice of using semicolons is technically the worse.
Therefore the direction we should be going is clear.
I believe it's a fair requirement for anyone adopting colored underline
to support colons too, and it's a reasonable move from applications to
slightly push the world forward, force developers to catch up with the
recent changes, that is: 1) recognize and at least ignore
colon-delimited parameters even if they aren't supported, 2) recognize
and support colon wherever they support the nonstandard semicolon
instead.
Should you come across any terminal emulator that supports 58 with
semicolons but not with colons, I think the cleanest you can do is
report a bug against them and ignore the problem; they should fix it.
It's yet another common misunderstanding that the truecolor syntax is
`38`/`48`/`58` followed by `:2:rrr:ggg:bbb`. The wording of T.416 is
terrible, but if you read carefully, there's another parameter of
color-space-id preceding the three color channels. Assuming you don't
care about color-space-id, the syntax is `38`/`48`/`58` followed by
`:2::rrr:ggg:bbb` and of course the trailing `m`.
This is only for true-color, the 256-color format doesn't have such
a parameter, it's `38`/`48`/`58` followed by `:5:index` and the final
`m`.