It is a design goal of extmarks that they allow precise tracking
of changes across undo/redo, including restore the exact positions
after a do/undo or undo/redo cycle. However this behavior is not useful
for all usecases. Many plugins won't keep marks around for long after
text changes, but uses them more like a cache until some external source
(like LSP semantic highlights) has fully updated to changed text and
then will explicitly readjust/replace extmarks as needed.
Add a "undo_restore" flag which is true by default (matches existing
behavior) but can be set to false to opt-out of this behavior.
Delete dead u_extmark_set() code.
regext_bt.c and regexp_nfa.c are inlined into regexp.c instead of
included as a header. This makes developer tools like clang-tidy and
clangd be able to understand the code better.
- silence false warnings on MSVC
- merge `clang-tidy` cmake target into `lintc` and remove the
corresponding make target
- use cmake's built-in endianness detection
Problem: complete_info() skips entries with 'noselect'
Solution: Check, if first entry is at original text state
Unfortunately, Commit daef8c74375141974d61b85199b383017644978c
introduced a regression, that when ':set completeopt+=noselect' is set
and no completion item has been selected yet, it did not fill the
complete_info['items'] list.
This happened, because the current match item did not have the
CP_ORIGINAL_TEXT flag set and then the cp->prev pointer did point to the
original flag item, which caused the following while loop to not being
run but being skipped instead.
So when the 'noselect' is set, only start with to the previous selection
item, if the initial completion item has the CP_ORIGINAL_TEXT flag set,
else use the 2nd previous item instead.
fixes: vim/vim#13451closes: vim/vim#1345257f9ce1a09
Co-authored-by: Christian Brabandt <cb@256bit.org>
refactor: use a more idiomatic loop to iterate over the cells
There are two cases in which the following assertion would fail:
```c
assert(g->icell < g->ncells);
```
1. If `g->ncells = 0`. Update this to be legal.
2. If an EOF is reached while parsing `wrap`. In this case, the unpacker
attempts to resume from `cells`, which is a bug. Create a new state
for parsing `wrap`.
Reference: https://neovim.io/doc/user/ui.html#ui-event-grid_line
Previously, 'rightleftcmd' was implemented by having all code which
would affect msg_col or output screen cells be conditional on `cmdmsg_rl`.
This change removes all that and instead implements rightleft as a
mirroring post-processing step.
In cases where the generated files depend on changes to Nvim itself,
generating the files with an older version of Nvim will fail because
those changes are not present in the older version.
For example, if a new option is added then the generator script should
be run with the version of Nvim that contains the new option, or else
the generation will fail.
Co-authored-by: dundargoc <gocdundar@gmail.com>
connection from any channel or stdio will unblock
remote_ui_wait_for_attach. Wait on stdio only if
only —embed specified, if both —embed and
—listen then wait on any channel.
Problem: `os_doskip` seems to be unnecessary since everything that sets it to true also returns an error, and `errmsg` being non-`NULL` already skips most of the processing.
Solution: Remove `os_doskip`.
While the interfaces for setting number and boolean options are now unified by #25394, there is still a separate `set_string_option` function that is used for setting a string option. This PR removes that function and merges it with set_option.
BREAKING CHANGE: `v:option_old` is now the old global value for all global-local options, instead of just string global-local options. Local value for a global-local number/boolean option is now unset when the option is set (e.g. using `:set` or `nvim_set_option_value`) without a scope, which means they now behave the same way as string options.
Ref: #25672
runtime(doc): all secure options should note this restriction in the documentation (vim/vim#13448)
Problem: Not all secure options document their status
Solution: Describe secure context :set restrictions in each help entry
8ebdbc9e6d
Co-authored-by: dkearns <dougkearns@gmail.com>
Problem: With 'smoothscroll' set, "w_skipcol" is not reset when unsetting 'wrap'.
Resulting in incorrect calculation of the cursor position.
Solution: Reset "w_skipcol" when unsetting 'wrap'.
fixes: vim/vim#12970closes: vim/vim#134391bf1bf569b
Problem: TextChangedI may not always trigger
Solution: trigger it in more cases: for insert/
append/change operations, and when
opening a new line,
fixes: vim/vim#13367closes: vim/vim#133754bca4897a1
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: [security] disallow setting env in restricted mode
Solution: Setting environment variables in restricted mode could
potentially be used to execute shell commands. Disallow this.
restricted mode: disable allow setting of environment variables
Setting environment variables in restricted mode, may have some unwanted
consequences. So, for example by setting $GCONV_PATH in restricted mode
and then calling the iconv() function, one may be able to execute some
unwanted payload, because the `iconv_open()` function internally uses
the `$GCONV_PATH` variable to find its conversion data.
So let's disable setting environment variables, even so this is no
complete protection, since we are not clearing the existing environment.
I tried a few ways but wasn't successful :(
One could also argue to disable the iconv() function completely in
restricted mode, but who knows what other API functions can be
influenced by setting some other unrelated environment variables.
So let's leave it as it is currently.
closes: vim/vim#13394
See: https://huntr.com/bounties/b0a2eda1-459c-4e36-98e6-0cc7d7faccfe/6b89dd6a72
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: Completion menu may be wrong
Solution: Check for the original direction of the completion menu,
add more tests, make it work with 'noselect'
completion: move in right direction when filling completion_info()
When moving through the insert completion menu and switching directions,
we need to make sure we start at the correct position in the list and
move correctly forward/backwards through it, so that we do not skip
entries and the selected item points to the correct entry in the list
of completion entries generated by the completion_info() function.
The general case is this:
1) CTRL-X CTRL-N, we will traverse the list starting from
compl_first_match and then go forwards (using the cp->next pointer)
through the list (skipping the very first entry, which has the
CP_ORIGINAL_TEXT flag set (since that is the empty/non-selected entry
2) CTRL-X CTRL-P, we will traverse the list starting from
compl_first_match (which now points to the last entry). The previous
entry will have the CP_ORIGINAL_TEXT flag set, so we need to start
traversing the list from the second prev pointer.
There are in fact 2 special cases after starting the completion menu
with CTRL-X:
3) CTRL-N and then going backwards by pressing CTRL-P again.
compl_first_match will point to the same entry as in step 1 above,
but since compl_dir_foward() has been switched by pressing CTRL-P
to backwards we need to pretend to be in still in case 1 and still
traverse the list in forward direction using the cp_next pointer
4) CTRL-P and then going forwards by pressing CTRL-N again.
compl_first_match will point to the same entry as in step 2 above,
but since compl_dir_foward() has been switched by pressing CTRL-N
to forwards we need to pretend to be in still in case 2 and still
traverse the list in backward direction using the cp_prev pointer
For the 'noselect' case however, this is slightly different again. When
going backwards, we only need to go one cp_prev pointer back. And
resting of the direction works again slightly different. So we need to
take the noselect option into account when deciding in which direction
to iterate through the list of matches.
related: vim/vim#13402
related: vim/vim#12971closes: vim/vim#13408daef8c7437
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: [security] overflow in :history
Solution: Check that value fits into int
The get_list_range() function, used to parse numbers for the :history
and :clist command internally uses long variables to store the numbers.
However function arguments are integer pointers, which can then
overflow.
Check that the return value from the vim_str2nr() function is not larger
than INT_MAX and if yes, bail out with an error. I guess nobody uses a
cmdline/clist history that needs so many entries... (famous last words).
It is only a moderate vulnerability, so impact should be low.
Github Advisory:
https://github.com/vim/vim/security/advisories/GHSA-q22m-h7m2-9mgm9198c1f2b1
N/A patch:
vim-patch:9.0.2073: typo in quickfix.c comments
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem:
`win_get_bordertext_col` returns column < 1 for right or center
aligned text, if its length is more than window width.
Solution:
Return max(resulting_column, 1)
Problem: cannot use buffer-number for errorformat
Solution: add support for parsing a buffer number using '%b' in
'errorformat'
closes: vim/vim#13419b731800522
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Problem: EXPAND flag set for filetype option
Solution: Remove P_EXPAND flag from the 'filetype' option
Remove P_EXPAND flag from the 'filetype' option
Problem: P_EXPAND flag is erroneously set for 'filetype' option
Solution: Remove the P_EXPAND flag
This flag is used by string options that accept file path values. See
:help set_env.
This appears to have been included in d5e8c92 and resulted from an
incorrect implementation of 'filetype' completion.
See vim/vim#12900 for a small discussion.
related: vim/vim#12900closes: vim/vim#134163932072ab4
Co-authored-by: Doug Kearns <dougkearns@gmail.com>
Problem: #25716 removed the logic for getting the correct value of the `'modified'` option when using `nvim_buf_get_option()`. So the function now returns an incorrect value of `'modified'` for scratch buffers.
Solution: Re-add the logic for getting the correct value of `'modified'`. Also make it so that the logic only exists in one place instead of being duplicated across multiple places in the code.
Problem: With lots of events, UI flush may be delayed for too long.
Solution: Flush UI when no input is available. This still flushes less
frequently than before #25629, when a flush also happens when
typeahead buffer is empty but input is available.
`SREQ_*` values are now actual typedef'd enums. `get_option_value_strict()` has also been refactored and split into two functions, `get_option_attrs()` for getting the option attributes, and `get_option_value_strict()` for getting the actual value. Moreover, it now returns an `OptVal`. Other miscellaneous refactors have also been made.
Problem: no digraph for quadruple prime
Solution: add quadruple prime digraph using 4'
closes: vim/vim#1338047416d1a74
Co-authored-by: Jonathan Wright <quaggy@gmail.com>
Problem: Vim9: crash with deferred function call and exception
Solution: Save and restore exception state
Crash when a deferred function is called after an exception and another
exception is thrown
closes: vim/vim#13376closes: vim/vim#13377c59c1e0d88
The change in check_due_timer() is N/A as Nvim calls timer callbacks on
the main loop.
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Problem: trim(): hard to use default mask (partly revert v9.0.2040)
Solution: use default mask when it is empty
The default 'mask' value is pretty complex, as it includes many
characters. Yet, if one needs to specify the trimming direction, the
third argument, 'trim()' currently requires the 'mask' value to be
provided explicitly.
Currently, an empty 'mask' will make 'trim()' call return 'text' value
that is passed in unmodified. It is unlikely that someone is using it,
so the chances of scripts being broken by this change are low.
Also, this reverts commit 9.0.2040 (which uses v:none for the default
and requires to use an empty string instead).
closes: vim/vim#133588079917447
vim-patch:9.0.2040: trim(): hard to use default mask
Problem: trim(): hard to use default mask
Solution: Use default 'mask' when it is v:none
The default 'mask' value is pretty complex, as it includes many
characters. Yet, if one needs to specify the trimming direction, the
third argument, 'trim()' currently requires the 'mask' value to be
provided explicitly.
'v:none' is already used to mean "use the default argument value" in
user defined functions. See |none-function_argument| in help.
closes: vim/vim#133636e6386716f
Co-authored-by: Illia Bobyr <illia.bobyr@gmail.com>
