Before, we needed to always pack an entire msgpack_rpc Object to
a continous memory buffer before sending it out to a channel.
But this is generally wasteful. it is better to just flush
whatever is in the buffer and then continue packing to a new buffer.
This is also done for the UI event packer where there are some extra logic
to "finish" of an existing batch of nevents/ncalls. This doesn't really
stop us from flushing the buffer, just that we need to update the state
machine accordingly so the next call to prepare_call() always will
start with a new event (even though the buffer might contain overflow
data from a large event).
Then we can just load metadata in C as a single msgpack blob. Which also
can be used directly as binarly data, instead of first unpacking all the
functions and ui_events metadata to immediately pack it again, which was
a bit of a silly walk (and one extra usecase of `msgpack_rpc_from_object`
which will get yak shaved in the next PR)
Problem:
The documentation flow (`gen_vimdoc.py`) has several issues:
- it's not very versatile
- depends on doxygen
- doesn't work well with Lua code as it requires an awkward filter script to convert it into pseudo-C.
- The intermediate XML files and filters makes it too much like a rube goldberg machine.
Solution:
Re-implement the flow using Lua, LPEG and treesitter.
- `gen_vimdoc.py` is now replaced with `gen_vimdoc.lua` and replicates a portion of the logic.
- `lua2dox.lua` is gone!
- No more XML files.
- Doxygen is now longer used and instead we now use:
- LPEG for comment parsing (see `scripts/luacats_grammar.lua` and `scripts/cdoc_grammar.lua`).
- LPEG for C parsing (see `scripts/cdoc_parser.lua`)
- Lua patterns for Lua parsing (see `scripts/luacats_parser.lua`).
- Treesitter for Markdown parsing (see `scripts/text_utils.lua`).
- The generated `runtime/doc/*.mpack` files have been removed.
- `scripts/gen_eval_files.lua` now instead uses `scripts/cdoc_parser.lua` directly.
- Text wrapping is implemented in `scripts/text_utils.lua` and appears to produce more consistent results (the main contributer to the diff of this change).
As only a few API functions make use of explicit freeing of the return
value, make it opt-in instead. The arena is always present under the
hood, so `Arena *arena` arg now doesn't mean anything other than getting
access to this arena. Also it is in principle possible to return an
allocated value while still using the arena as scratch space for other
stuff (unlikely, but there no reason to not allow it).
Note: this contains two _temporary_ changes which can be reverted
once the Arena vs no-Arena distinction in API wrappers has been removed.
Both nlua_push_Object and object_to_vim_take_luaref() has been changed
to take the object argument as a pointer. This is not going to be
necessary once these are only used with arena (or not at all) allocated
Objects.
The object_to_vim() variant which leaves luaref untouched might need to
stay for a little longer.
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.
Currently having two separate memory strategies for API return values is
a bit unnecessary, and mostly a consequence of converting the hot spot
cases which needed it first. But there is really no downside to using
arena everywhere (which implies also directly using strings which are
allocated earlier or even statically, without copy).
There only restriction is we need to know the size of arrays in advance,
but this info can often be passed on from some earlier stage if it is
missing.
This collects some "small" cases. The more complex stuff will get a PR
each.
Implement api_keydict_to_dict as the complement to api_dict_to_keydict
Fix a conversion error when nvim_get_win_config gets called from lua,
where Float values "x" and "y" didn't get converted to lua numbers.
"export" only prevents IWYU from adding these headers if the headers
that export them are included, while "private" ensures that IWYU never
adds these headers.
It isn't really useful to put anonymous enums only used as arguments to
functions calls in _defs.h headers, as they will only be used by a file
that calls those functions, which requires including a non-defs header.
Also move os_msg() and os_errmsg() back to message.h, as on Windows they
are actual functions instead of macros.
Also remove gettext.h and globals.h from private/helpers.h.
Problem:
Not all Lua code is checked by stylua. Automating code-style is an
important mechanism for reducing time spent on accidental
(non-essential) complexity.
Solution:
- Enable lintlua for `src/` directory.
followup to 517f0cc634
FUNC_ATTR_* should only be used in .c files with generated headers.
Defining FUNC_ATTR_* as empty in headers causes misuses of them to be
silently ignored. Instead don't define them by default, and only define
them as empty after a .c file has included its generated header.
We already have an extensive suite of static analysis tools we use,
which causes a fair bit of redundancy as we get duplicate warnings. PVS
is also prone to give false warnings which creates a lot of work to
identify and disable.
Problem: some API functions that check textlock (usually those that can change
curwin or curbuf) can break the cmdwin.
Solution: make FUNC_API_CHECK_TEXTLOCK call text_locked() instead, which already
checks for textlock, cmdwin and `<expr>` status.
Add FUNC_API_TEXTLOCK_ALLOW_CMDWIN to allow such functions to be usable in the
cmdwin if they can work properly there; the opt-in nature of this attribute
should hopefully help mitigate future bugs.
Also fix a regression in #22634 that made functions checking textlock usable in
`<expr>` mappings, and rename FUNC_API_CHECK_TEXTLOCK to FUNC_API_TEXTLOCK.
problem: can we have Serde?
solution: we have Serde at home
This by itself is just a change of notation, that could be quickly
merged to avoid messy merge conflicts, but upcoming changes are planned:
- keysets no longer need to be defined in one single file. `keysets.h` is
just the initial automatic conversion of the previous `keysets.lua`.
keysets just used in a single api/{scope}.h can be moved to that file, later on.
- Typed dicts will have more specific types than Object. this will
enable most of the existing manual typechecking boilerplate to be eliminated.
We will need some annotation for missing value, i e a boolean will
need to be represented as a TriState (none/false/true) in some cases.
- Eventually: optional parameters in form of a `Dict opts` final
parameter will get added in some form to metadata. this will require
a discussion/desicion about type forward compatibility.
Problem:
Build is not reproducible, because generated source files (.c/.h/) are not
deterministic, mostly because Lua pairs() is unordered by design (for security).
https://github.com/LuaJIT/LuaJIT/issues/626#issuecomment-707005671https://www.lua.org/manual/5.1/manual.html#pdf-next
> The order in which the indices are enumerated is not specified [...]
>
>> The hardening of the VM deliberately randomizes string hashes. This in
>> turn randomizes the iteration order of tables with string keys.
Solution:
- Update the code generation scripts to be deterministic.
- That is only a partial solution: the exported function
(funcs_metadata.generated.h) and ui event
(ui_events_metadata.generated.h) metadata have some mpack'ed
tables, which are not serialized deterministically.
- As a workaround, introduce `PRG_GEN_LUA` cmake setting, so you can
inject a modified build of luajit (with LUAJIT_SECURITY_PRN=0)
that preserves table order.
- Longer-term we should change the mpack'ed data structure so it no
longer uses tables keyed by strings.
Closes#20124
Co-Authored-By: dundargoc <gocdundar@gmail.com>
Co-Authored-By: Arnout Engelen <arnout@bzzt.net>
Allow Include What You Use to remove unnecessary includes and only
include what is necessary. This helps with reducing compilation times
and makes it easier to visualise which dependencies are actually
required.
Work on https://github.com/neovim/neovim/issues/549, but doesn't close
it since this only works fully for .c files and not headers.
In case nvim A sends nvim_error_event to nvim B, it would
respond with another nvim_error_event due to unknown
request name. Fix this by adding dummy request handler for now.
Problem: Get ml_get error when deleting a line in 'completefunc'. (Yegappan
Lakshmanan)
Solution: Lock the text while evaluating 'completefunc'.
ff06f283e3
Fix a mistake in the porting of patch 8.1.0098.
Cherry-pick Test_run_excmd_with_text_locked() from patch 8.2.0270.
Cherry-pick test_gf.vim changes from patch 8.2.0369.
Cherry-pick message change from later patches.
Problem:
1. The main log routine does not protect itself against recursion.
log_lock() doesn't guard against recursion, it would deadlock...
2. 22b52dd462 (#11501) regressed 6f27f5ef91 (#10172), because
set_init_1..process_spawn tries to log (see backtrace below), but the
mutex isn't initialized yet. Even if the mutex were valid, we don't
want early logging to fallback to stderr because that can break
embedders when stdio is used for RPC.
frame 1: 0x00000001001d54f4 nvim`open_log_file at log.c:205:7
frame 2: 0x00000001001d5390 nvim`logmsg(log_level=1, context="UI: ", func_name=0x0000000000000000, line_num=-1, eol=true, fmt="win_viewport") at log.c:150:20
frame : 0x000000010039aea2 nvim`ui_call_win_viewport(grid=2, win=1000, topline=0, botline=1, curline=0, curcol=0, line_count=1) at ui_events_call.generated.h:321:3
frame 4: 0x00000001003dfefc nvim`ui_ext_win_viewport(wp=0x0000000101816400) at window.c:939:5
frame 5: 0x00000001003ec5b4 nvim`win_ui_flush at window.c:7303:7
frame 6: 0x00000001003a04c0 nvim`ui_flush at ui.c:508:3
frame 7: 0x00000001002966ba nvim`do_os_system(argv=0x0000600000c0c000, input=0x0000000000000000, len=0, output=0x0000000000000000, nread=0x00007ff7bfefe830, silent=false, forward_output=false) at shell.c:894:3
frame 8: 0x0000000100295f68 nvim`os_call_shell(cmd="unset nonomatch; vimglob() { while [ $# -ge 1 ]; do echo \"$1\"; shift; done }; vimglob >/var/folders/gk/3tttv_md06987tlwpyp62jrw0000gn/T/nvimwwvwfD/0 ~foo", opts=kShellOptExpand | kShellOptSilent | kShellOptHideMess, extra_args=0x0000000000000000) at shell.c:663:18
frame 9: 0x0000000100295845 nvim`call_shell(cmd="unset nonomatch; vimglob() { while [ $# -ge 1 ]; do echo \"$1\"; shift; done }; vimglob >/var/folders/gk/3tttv_md06987tlwpyp62jrw0000gn/T/nvimwwvwfD/0 ~foo", opts=kShellOptExpand | kShellOptSilent | kShellOptHideMess, extra_shell_arg=0x0000000000000000) at shell.c:712:14
frame 10: 0x0000000100294c6f nvim`os_expand_wildcards(num_pat=1, pat=0x00007ff7bfefeb20, num_file=0x00007ff7bfefee58, file=0x00007ff7bfefee60, flags=43) at shell.c:328:7
...
frame 23: 0x000000010028ccef nvim`expand_env_esc(srcp=",~foo", dst="~foo", dstlen=4094, esc=false, one=false, prefix=0x0000000000000000) at env.c:673:17
frame 24: 0x000000010026fdd5 nvim`option_expand(opt_idx=29, val=",~foo") at option.c:1950:3
frame 25: 0x000000010026f129 nvim`set_init_1(clean_arg=false) at option.c:558:19
frame 26: 0x00000001001ea25e nvim`early_init(paramp=0x00007ff7bfeff5f0) at main.c:198:3
frame 27: 0x00000001001ea6bf nvim`main(argc=1, argv=0x00007ff7bfeff848) at main.c:255:3
Solution:
1. Check for recursion, show "internal error" message.
- FUTURE: when "remote TUI" is merged, can we remove log_lock()?
2. Skip logging if log_init wasn't called yet.
Fixes#15147 and fixes#15497. Also sketch "subdir" caching. Currently
this only caches whether an rtp entry has a "lua/" subdir but we could
consider cache other subdirs potentially or even "lua/mybigplugin/"
possibly.
Note: the async_leftpad test doesn't actually fail on master, at least
not deterministically (even when disabling the fast_breakcheck
throttling). It's still useful as a regression test for further changes
and included as such.
Do not copy a lot of lua strings (dict keys) to just strequal() them
Just compare them directly to a dedicated hash function.
feat(generators): HASHY McHASHFACE
These functions do not involve msgpack. Initially the nvim api was
sometimes called the "msgpack API", but entry points from vim script
and lua are equally valid (and don't need to reference "msgpack")
Note: the reason for removing them is not that there after this refactor
is no use of them, but rather that having them available is an
anti-pattern: they manange an _extra_ heap allocation which has
nothing to do with the functionality of the map itself (khash
manages the real buffers internally). In case there happens to
be a reason to allocate the map structure itself later, this
should be made explicit using xcalloc/xfree calls.
