Problem:
After #37722 splice optimization for :packadd, nvim crashes with SIGSEGV
on startup while running `runtime! lua/xxx/*` and sourced file executes
multiple `:packadd`.
Solution:
While `do_in_cached_path` is executing, it doesn't expect reference to
runtime_search_path changes. But when callback is called, and
add_pack_dir_to_rtp does 'splice' it may trigger realloc, and change
address. Check runtime_search_path_ref to prevent ref held by
do_in_cached_path changes.
fixes#37586
when doing `packadd mypackage` up to two exact paths are added
to &rtp. Instead of recalculating runtime_search_path from scratch,
we can "just" splice these two paths in
This is simple in theory, but get complicated in practice as
"after" dirs do exist and need some wrangling.
Echasnovski did some benchmarking, to show that this reduces overhead
of a init.lua configuration style where separate `packadd!` calls are
used spread out during the config. In addition, "batched" addition
(either using "start" packages or packadd! a lot of opt packages at
once) does not regress.
A theoretical simplification could be to NEVER explicitly add "after"
dirs to &rtp, but implicitly add all existing "after" dirs in reverse
order when calculating the effective run time path. This might be tricky
to do without breaking 12 tpope plugins again tho.
We might also instead consider solutions where &rtp remains fully expanded but no longer is the main source of truth. But this is all post 0.12 work. This PR is an alright stopgap to make 0.12 fully support intended use cases of vim.pack.add() .
Problem: Not always using GA_CONCAT_LITERAL with string literals.
(after: v9.2.0031)
Solution: Use the GA_CONCAT_LITERAL, instead of ga_concat_len.
(John Marriott)
closes: vim/vim#19468fc90d8087a
Co-authored-by: John Marriott <basilisk@internode.on.net>
Problem: Newlines intended to write messages below the cmdline or to
mark the start of a new message on message grid are emitted
through ext_messages. This results in unnecessary newlines for
a UI that has decoupled its message area from the cmdline.
msg_col is set directly in some places which is not transmitted
to msg_show events.
Various missing message kind for list commands.
Trailing newlines on various list commands.
Solution: Only emit such newlines without ext_messages enabled.
Use msg_advance() instead of setting msg_col directly.
Assign them the "list_cmd" kind.
Ensure no trailing newline is printed.
Problem:
Can't use `:source` to run a Lua codeblock (treesitter injection) in
a help (vimdoc) file.
Solution:
Use treesitter to parse the range and treat it as Lua if detected as
such.
Problem: Exe stack length can be wrong without being detected.
Solution: Add a check when ABORT_ON_INTERNAL_ERROR is defined.
e31ee86859
vim-patch:8.2.3262: build failure when ABORT_ON_INTERNAL_ERROR is defined
Port patch 9.0.1454 for "make formatc".
Co-authored-by: Bram Moolenaar <Bram@vim.org>
Problem: Vim9: class_name definition can be improved
Solution: Use string_T to store class_name, avoid using STRLEN() calls,
simplify code, remove unused definition of struct
oc_newmember_S (John Marriott)
Use struct string_T to store the field class_name in struct class_T,
which means we can just use the .length field in struct string_T
instead of measuring it.
In addition:
1. In eval.c use string_T to store class_name and s in function
class_tv2string().
2. In vim9type.c change some calls from ga_concat() to ga_concat_len()
where the length is known.
3. In vim9class.c remove unused struct definition oc_newmember_S.
Change some calls from ga_concat() to ga_concat_len() where the
length is known.
4. In scriptfile.c use string_T to store type_name, class_name and
es_name in function estack_sfile().
5. In function estack_sfile() simplify construction of the grow array ga
and change some calls from ga_concat() to ga_concat_len() when the
length is known.
closes: vim/vim#189252019321e0b
Co-authored-by: John Marriott <basilisk@internode.on.net>
Problem: vim9 no class identifiers in stack dumps
Solution: Prefix class members in stack traces with the class name
followed by a dot.
closes: vim/vim#12866closes: vim/vim#120780ffc17aa47
Co-authored-by: LemonBoy <thatlemon@gmail.com>
Problem: MS-Windows: Relative import in a script sourced from a buffer
doesn't work (Ernie Rael)
Solution: Set a filename, so that we are not trying to use
script-relative filename (Yegappan Lakshmanan)
When a script is sourced from a buffer, the file name is set to ":source
buffer=". In MS-Windows, the ":" is a path separator character (used
after a drive letter). This results in the code trying to use the ":"
prefix to import the script on MS-Windows. To fix this, when importing a
script from a script sourced from a buffer with nofile, don't use
a script relative path name.
fixesvim/vim#14588closes: vim/vim#14603f135fa28e4
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
These are not needed after #35129 but making uncrustify still play nice
with them was a bit tricky.
Unfortunately `uncrustify --update-config-with-doc` breaks strings
with backslashes. This issue has been reported upstream,
and in the meanwhile auto-update on every single run has been disabled.
Problem:
Coverity reports string null termination issue:
*** CID 569464: Memory - string handling (STRING_NULL)
/src/nvim/runtime.c: 1374 in ExpandRTDir_int()
1372 if (flags & DIP_START) {
1373 memcpy(tail - 15, "pack/*/start/*/", 15); // NOLINT
>>> CID 569464: Memory - string handling (STRING_NULL)
>>> Passing unterminated string "tail - 15" to "globpath", which expects a null-terminated string.
1374 globpath(p_pp, tail - 15, gap, glob_flags, expand_dirs);
Similar issues occur at lines 1376, 1381, and 1383 where memcpy()
constructs strings passed to globpath() without null termination.
Solution:
Replace dangerous pointer arithmetic and memcpy() with direct snprintf()
construction of complete search paths. This eliminates the need for
buffer reuse through pointer offsets and ensures all strings passed to
globpath() are properly null-terminated.
vim-patch:9.1.1515: Coverity complains about potential unterminated strings
Problem:
vim.uv.os_setenv gets "stuck" per-key. #32550
Caused by the internal `envmap` cache. #7920
:echo $FOO <-- prints nothing
:lua vim.uv.os_setenv("FOO", "bar")
:echo $FOO <-- prints bar (as expected)
:lua vim.uv.os_setenv("FOO", "fizz")
:echo $FOO <-- prints bar, still (not expected. Should be "fizz")
:lua vim.uv.os_unsetenv("FOO")
:echo $FOO <-- prints bar, still (not expected. Should be nothing)
:lua vim.uv.os_setenv("FOO", "buzz")
:echo $FOO <-- prints bar, still (not expected. Should be "buzz")
Solution:
- Remove the `envmap` cache.
- Callers to `os_getenv` must free the result.
- Update all call-sites.
- Introduce `os_getenv_noalloc`.
- Extend `os_env_exists()` the `nonempty` parameter.
Problem: ":so" command may read after end of buffer.
Solution: Compute length of text properly.
4748c4bd64
Co-authored-by: Bram Moolenaar <Bram@vim.org>
Problem: Sourcing buffer lines is too complicated.
Solution: Simplify the code. Make it possible to source Vim9 script lines.
(Yegappan Lakshmanan, closesvim/vim#9974)
85b43c6cb7
This commit changes the behavior of sourcing buffer lines to always have
a script ID, although sourcing the same buffer always produces the same
script ID.
vim-patch:9.1.0372: Calling CLEAR_FIELD() on the same struct twice
Problem: Calling CLEAR_FIELD() on the same struct twice.
Solution: Remove the second CLEAR_FIELD(). Move the assignment of
cookie.sourceing_lnum (zeertzjq).
closes: vim/vim#14627f68517c167
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Problem: Need to write script to a file to be able to source them.
Solution: Make ":source" use lines from the current buffer. (Yegappan
Lakshmanan et al., closesvim/vim#9967)
36a5b6867b
Most code and test changes are reverted or modified again in patch
8.2.4603, so only port parts that are untouched in patch 8.2.4603.
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Problem: When setting an option, mapping etc. from Lua without -V1, the
script ID is set to SID_LUA even if there already is a script
ID assigned by :source.
Solution: Don't set script ID to SID_LUA if it is already a Lua script.
Also add _editor.lua to ignorelist to make script context more
useful when using vim.cmd().
Problem: When calling an API from Vimscript to set an option, mapping,
etc., :verbose shows that it's set from an API client.
Solution: Don't override current_sctx.sc_sid when calling an API from
Vimscript. Also fix the inverse case where API channel id is
not set when calling an API from RPC. Move channel id into
sctx_T to make saving and restoring easier.
Related #8329
Problem: Highlight group id is not propagated to the end of the message call
stack, where ext_messages are emitted.
Solution: Refactor message functions to pass along highlight group id
instead of attr id.
Problem:
If $NVIM_APPNAME is a relative dir path, Nvim fails to start its
primary/default server, and `v:servername` is empty.
Root cause is d34c64e342, but this wasn't
noticed until 96128a5076 started reporting the error more loudly.
Solution:
- `server_address_new`: replace slashes "/" in the appname before using
it as a servername.
- `vim_mktempdir`: always prefer the system-wide top-level "nvim.user/"
directory. That isn't intended to be specific to NVIM_APPNAME; rather,
each *subdirectory* ("nvim.user/xxx") is owned by each Nvim instance.
Nvim "apps" can be identified by the server socket(s) stored in those
per-Nvim subdirs.
fix#30256
Problem: Cannot use an autoload function from a package under start.
Solution: Also look in the "start" package directory. (Bjorn Linse,
closesvim/vim#7193)
223a950a85
Nvim already does this in do_in_cached_path(), and this change has no
effect in Nvim as Nvim removes DIP_START after do_in_cached_path().
Accidentally failed to mark as ported:
vim-patch:8.2.1731: Vim9: cannot use += to append to empty NULL list
Co-authored-by: bfredl <bjorn.linse@gmail.com>
The latter was mostly relevant with the past char_u madness.
NOTE: STRCAT also functioned as a counterfeit "NOLINT" for clint
apparently. But NOLINT-ing every usecase is just the same as disabling
the check entirely.
Problem: Vim9: missing :endif not reported when using :windo.
Solution: Pass a getline function to do_cmdline(). (closesvim/vim#7650)
9567efa1b4
Co-authored-by: Bram Moolenaar <Bram@vim.org>
Problem: Using xstrlcpy() when the exact length of the string to be
copied is known is not ideal because it requires adding 1 to
the length and an unnecessary strlen().
Solution: Add xmemcpyz() and use it in place of such xstrlcpy() calls.
Problem: POSIX function name in exarg struct causes issues
on OpenVMS
Solution: Rename getline member in exarg struct to ea_getline,
remove isinf() workaround for VMS
There are compilers that do not treat well POSIX functions - like
getline - usage in the structs.
Older VMS compilers could digest this... but the newer OpenVMS compilers
( like VSI C x86-64 X7.4-843 (GEM 50XB9) ) cannot deal with these
structs. This could be limited to getline() that is defined via
getdelim() and might not affect all POSIX functions in general - but
avoiding POSIX function names usage in the structs is a "safe side"
practice without compromising the functionality or the code readability.
The previous OpenVMS X86 port used a workaround limiting the compiler
capabilities using __CRTL_VER_OVERRIDE=80400000
In order to make the OpenVMS port future proof, this pull request
proposes a possible solution.
closes: vim/vim#137046fdb628082
Co-authored-by: Zoltan Arpadffy <zoltan.arpadffy@gmail.com>
The way ml_replace_buf is implemented makes it unfriendly for
being used in a loop: every call allocates a scratch buffer for putting
the line into the "dirty" state. This then immediately needs to be freed
as the next ml_replace_buf and/or ml_append_buf call will flush that buffer.
It's better to later pay the price of allocating the scratch buffer only if
the line is being immediately edited (likely when using the API to only
change one line) with an extra memcpy, than allocating that buffer
multiple times every time the API is called.
Of course, a separate xmalloc/xfree cycle for each time the dirty line
changes is unwanted to begin with. But fixing that is a later refactor.
and for return value of nlua_exec/nlua_call_ref, as this uses
the same family of functions.
NB: the handling of luaref:s is a bit of a mess.
add api_luarefs_free_XX functions as a stop-gap as refactoring
luarefs is a can of worms for another PR:s.
as a minor feature/bug-fix, nvim_buf_call and nvim_win_call now preserves
arbitrary return values.
Remove `export` pramgas from defs headers as it causes IWYU to believe
that the definitions from the defs headers comes from main header, which
is not what we really want.
