| Age | Commit message (Collapse) | Author |
|---|
|
Problem: Left shift is incorrect with vartabstop and shiftwidth=0
Solution: make tabstop_at() function aware of shift direction
(Gary Johnson)
The problem was that with 'vartabstop' set and 'shiftwidth' equal 0,
left shifts using << were shifting the line to the wrong column. The
tabstop to the right of the first character in the line was being used
as the shift amount instead of the tabstop to the left of that first
character.
The reason was that the tabstop_at() function always returned the value
of the tabstop to the right of the given column and was not accounting
for the direction of the shift.
The solution was to make tabstop_at() aware of the direction of the
shift and to choose the tabtop accordingly.
A test was added to check this behavior and make sure it doesn't
regress.
While at it, also fix a few indentation/alignment issues.
fixes: #14864
closes: #14887
Signed-off-by: Gary Johnson <garyjohn@spocom.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: MS-Windows: Compiler warnings
Solution: Resolve size_t to int warnings
closes: #14874
A couple of warnings in ex_docmd.c have been resolved by modifying their
function argument types, followed by some changes in various function
call sites. This also allowed removal of some casts to cope with
size_t/int conversion.
Signed-off-by: Mike Williams <mrmrdubya@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: getregionpos() behaves inconsistently for a partly-selected
multibyte char.
Solution: Always use column of the first byte for a partly-selected
multibyte char (zeertzjq).
closes: #14851
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Can't use a blockwise selection with a width for getregion().
Solution: Add support for blockwise selection with width like the return
value of getregtype() or the "regtype" value of TextYankPost
(zeertzjq).
closes: #14842
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: getregionpos() can't properly indicate positions beyond eol.
Solution: Add an "eol" flag that enables handling positions beyond end
of line like getpos() does (zeertzjq).
Also fix the problem that a position still has the coladd beyond the end
of the line when its column has been clamped. In the last test case
with TABs at the end of the line the old behavior is obviously wrong.
I decided to gate this behind a flag because returning positions that
don't correspond to actual characters in the line may lead to mistakes
for callers that want to calculate the length of the selected text, so
the behavior is only enabled if the caller wants it.
closes: #14838
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Wrong yanking with exclusive selection and virtualedit=all,
and integer overflow when using getregion() on it.
Solution: Set coladd when decreasing column and 'virtualedit' is active.
Add more tests for getregion() with 'virtualedit' (zeertzjq).
closes: #14830
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: getregionpos() doesn't handle one char selection.
Solution: Handle startspaces differently when is_oneChar is set.
Also add a test for an exclusive charwise selection with
multibyte chars (zeertzjq)
closes: #14825
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: too many strlen() calls in search.c
Solution: refactor code and remove more strlen() calls,
use explicit variable to remember strlen
(John Marriott)
closes: #14796
Signed-off-by: John Marriott <basilisk@internode.on.net>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: getregionpos() wrong with blockwise mode and multibyte.
Solution: Use textcol and textlen instead of start_vcol and end_vcol.
Handle coladd properly (zeertzjq).
Also remove unnecessary buflist_findnr() in add_regionpos_range(), as
getregionpos() has already switched buffer.
closes: #14805
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Cannot move to previous/next rare word
(Colin Kennedy)
Solution: Add the ]r and [r motions (Christ van Willegen)
fixes: #14773
closes: #14780
Signed-off-by: Christ van Willegen - van Noort <github.com@vanwillegen-vannoort.nl>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Some functions are not tested
Solution: Add a few more tests, fix a few minor problems
(Yegappan Lakshmanan)
closes: #14789
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: regionpos may leak memory on error, coverity
complains about dereferencing Null pointer
Solution: free all list pointers (after v9.1.394),
return early if buflist_findnr() returns NULL
closes: #14731
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Cannot get a list of positions describing a region
(Justin M. Keyes, after v9.1.0120)
Solution: Add the getregionpos() function
(Shougo Matsushita)
fixes: #14609
closes: #14617
Co-authored-by: Justin M. Keyes <justinkz@gmail.com>
Signed-off-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: cursor() and getregion() don't handle v:maxcol well.
Solution: Add special handling for v:maxcol like setpos() does.
(zeertzjq)
closes: #14698
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: null value tests not sufficient
Solution: Add a more comprehensive test for null values
(Yegappan Lakshmanan)
closes: #14701
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: crash with null_class and null_object
(Aliaksei Budavei)
Solution: Handle null_class and null_object correctly
(Yegappan Lakshmanan)
fixes: #14678
closes: #14681
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: console dialog cannot save unnamed buffers
Solution: set bufname before save (glepnir). Define dialog_con_gui
to test for GUI+Console dialog support, use it to skip
the test when the GUI feature has been defined.
Note: The dialog_changed() function will also try to call the
browse_save_fname() function, when FEAT_BROWSE is defined (which is only
defined in a GUI build of Vim). This will eventually lead to a call of
do_browse(), which causes an error message if a GUI is not currently
running (see the TODO: in do_browse()) and will then lead to a failure
in Test_goto_buf_with_onfirm().
Therefore, we must disable the Test_goto_buf_with_onfirm(), when the
dialog_con_gui feature is enabled (which basically means dialog feature
for GUI and Console builds, in contrast to the dialog_con and dialog_gui
feature).
(Previously this wasn't a problem, because the test aborted in the YES
case for the :confirm :b XgotoConf case and did therefore not run into
the browse function call)
closes: #14398
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Filetype may be undetected when a SwapExists autocommand sets
filetype in another buffer.
Solution: Make filetype detection state buffer-specific. Also fix a
similar problem for 'modified' (zeertzjq).
closes: #14344
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: No enum support
Solution: Implement enums for Vim9 script
(Yegappan Lakshmanan)
closes: #14224
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: too many STRLEN() calls
Solution: Make use of ml_get_len() calls instead
(John Marriott)
closes: #14123
Signed-off-by: John Marriott <basilisk@internode.on.net>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Internal error with blockwise getregion() in another buffer
Solution: Also change curwin->w_buffer when changing curbuf (zeertzjq)
closes: #14179
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Internal error or crash when passing invalid position to
getregion().
Solution: Give an error for invalid position (zeertzjq).
closes: #14172
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Duplicate assignment in f_getregion().
Solution: Remove the duplicate assignment. Also improve getregion()
docs wording and fix an unrelated typo (zeertzjq)
closes: #14154
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: can only call getregion() for current buffer
Solution: Allow to retrieve selections from different buffers
(Shougo Matsushita)
closes: #14131
Co-authored-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: can't call internal methods with objects
Solution: Add support for empty(), len() and string() function
calls for objects (Yegappan Lakshmanan)
closes: #14129
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: getregion() can be improved (after v9.1.120)
Solution: change getregion() implementation to use pos as lists and
one optional {opt} dictionary (Shougo Matsushita)
Note: The following is a breaking change!
Currently, the getregion() function (included as of patch v9.1.120) takes
3 arguments: the first 2 arguments are strings, describing a position,
arg3 is the type string.
However, that is slightly inflexible, there is no way to specify
additional arguments. So let's instead change the function signature to:
getregion(pos1, pos2 [, {Dict}]) where both pos1 and pos2 are lists.
This is slightly cleaner, and gives us the flexibility to specify
additional arguments as key/value pairs to the optional Dict arg.
Now it supports the "type" key to specify the selection type
(characterwise, blockwise or linewise) and now in addition one can also
define the selection type, independently of what the 'selection' option
actually is.
Technically, this is a breaking change, but since the getregion()
Vimscript function is still quite new, this should be fine.
closes: #14090
Co-authored-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Naming a non-pointer variable "oap" is strange.
Solution: Rename it to "oa". Also prevent using freed memory in case of
memory allocation failure. (zeertzjq)
closes: #14075
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Internal error when passing mark in another buffer to
getregion().
Solution: Don't allow marks in another buffer (zeertzjq)
closes: #14076
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Internal error when passing mark in another buffer to getregion()
|
|
Problem: Some minor issues with the getregion() function
Solution: Fix examples in the help, use OP_NOP op_type and MBLOCK
as motion_type in f_getreg(), update vim syntax to
for getregion() (Maxim Kim)
```
:xnoremap <CR>
\ <Cmd>echow getregion('v', '.', mode())<CR>
```
`echo` while in visual mode has no visible effect, thus people trying
example might be frustrated as it looks like nothing happens.
So the option is to change it to `echow` or `echom`.
With `echom` it is again has no visible effect but one can at least inspect `:messages`.
On the other hand `echow` showes selected text in a popup window.
```
Can also be used as a |method|: >
'.'->getregion("'a', 'v')
```
Here is the typo, which makes example invalid, should be `("'a", ...`
closes: #14064
Signed-off-by: Maxim Kim <habamax@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: hard to get visual region using Vim script
Solution: Add getregion() Vim script function
(Shougo Matsushita, Jakub Łuczyński)
closes: #13998
closes: #11579
Co-authored-by: =?UTF-8?q?Jakub=20=C5=81uczy=C5=84ski?= <doubleloop@o2.pl>
Co-authored-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Shougo Matsushita <Shougo.Matsu@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Not able to use diff() with 'diffexpr'
(rickhowe, after v9.1.0096)
Solution: Use a default context length of 0, update diff() help text,
add a test for using diff() with 'diffexpr'
(Yegappan Lakshmanan)
closes: #14013
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Need a diff() Vim script function
Solution: Add the diff() Vim script function using the
xdiff internal diff library, add support for
"unified" and "indices" mode.
(Yegappan Lakshmanan)
fixes: #4241
closes: #12321
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim is missing a foreach() func
Solution: Implement foreach({expr1}, {expr2}) function,
which applies {expr2} for each item in {expr1}
without changing it (Ernie Rael)
closes: #12166
Signed-off-by: Ernie Rael <errael@raelity.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Cannot easily get the list of matches
Solution: Add the matchstrlist() and matchbufline() Vim script
functions (Yegappan Lakshmanan)
closes: #13766
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: is*() and to*() function may be unsafe
Solution: Add SAFE_* macros and start using those instead
(Keith Thompson)
Use SAFE_() macros for is*() and to*() functions
The standard is*() and to*() functions declared in <ctype.h> have
undefined behavior for negative arguments other than EOF. If plain char
is signed, passing an unchecked value from argv for from user input
to one of these functions has undefined behavior.
Solution: Add SAFE_*() macros that cast the argument to unsigned char.
Most implementations behave sanely for negative arguments, and most
character values in practice are non-negative, but it's still best
to avoid undefined behavior.
The change from #13347 has been omitted, as this has already been
separately fixed in commit ac709e2fc0db6d31abb7da96f743c40956b60c3a
(v9.0.2054)
fixes: #13332
closes: #13347
Signed-off-by: Keith Thompson <Keith.S.Thompson@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: builtin funcs may accept a non-value
Solution: Restrict builtin functions that accept `type`
This PR finishes off detection and prevention of using a type as a
value. It takes care of builtin functions. However there are some
builtin functions, that need to be able to handle types as well as
non-args: instanceof(), type(), typename(), string().
A "bit", FE_X, is added to funcentry_T; when set, the builtin function
can handle a type (class or type-alias) in addition to a value.
Noteworthy change: Discovered that in compile_call() the builtin add()
is compiled inline instead of calling the builtin. Had to add a check
there.
closes: #13688
Signed-off-by: Ernie Rael <errael@raelity.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: can simplify arg type checking code
Solution: In `f_argcheck` array use `arg_any`, instead of NULL
closes: #13674
Signed-off-by: Ernie Rael <errael@raelity.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: instanceof() should use varargs as second arg
Solution: Modify `instanceof()` to use varargs instead of list
Modify `instanceof()` to use varargs instead of list
Valid `instanceof()` arguments are `type`s. A `type` is not a value;
it cannot be added to a list.
This change is non-compatible with the current usage of instanceof;
but instanceof is relatively new and it's a trivial change.
fixes: #13421
closes: #13644
Signed-off-by: Ernie Rael <errael@raelity.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: No test for mode() when executing Ex commands
Solution: Add some test cases and simplify several other test cases.
Also add a few more test cases for ModeChanged.
closes: #13588
Signed-off-by: zeertzjq <zeertzjq@outlook.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Problem with initializing the length of range() lists
Solution: Set length explicitly when it shouldn't contain any items
range() may cause a wrong calculation of list length, which may later
then cause a segfault in list_find(). This is usually not a problem,
because range_list_materialize() calculates the length, when it
materializes the list.
In addition, in list_find() when the length of the range was wrongly
initialized, it may seem to be valid, so the check for list index
out-of-bounds will not be true, because it is called before the list is
actually materialized. And so we may eventually try to access a null
pointer, causing a segfault.
So this patch does 3 things:
- In f_range(), when we know that the list should be empty, explicitly
set the list->lv_len value to zero. This should happen, when
start is larger than end (in case the stride is positive) or
end is larger than start when the stride is negative.
This should fix the underlying issue properly. However,
- as a safety measure, let's check that the requested index is not
out of range one more time, after the list has been materialized
and return NULL in case it suddenly is.
- add a few more tests to verify the behaviour.
fixes: #13557
closes: #13563
Co-authored-by: Tim Pope <tpope@github.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: No support for type aliases
Solution: Implement :type command
A type definition is giving a name to a type specification. This also known
type alias.
:type ListOfStrings = list<string>
The type alias can be used wherever a built-in type can be used. The type
alias name must start with an upper case character.
closes: #13407
Signed-off-by: Christian Brabandt <cb@256bit.org>
Signed-off-by: Yegappan Lakshmanan <yegappan@yahoo.com>
|
|
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: #13394
See: https://huntr.com/bounties/b0a2eda1-459c-4e36-98e6-0cc7d7faccfe/
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9: islocked() needs more work
Solution: rework islocked() and remove sync_root
from get_lval()
closes: #13329
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Ernie Rael <errael@raelity.com>
|
|
Problem: Vim9: no support for funcrefs
Solution: Add support for object/class funcref members
closes: #11981 #12417 #12960 #12324 #13333
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
|
|
Problem: Vim9: does not handle islocked() from a method correctly
Solution: Handle islocked() builtin from a method.
- Setup `lval_root` from `f_islocked()`.
- Add function `fill_exec_lval_root()` to get info about executing method.
- `sync_root` added in get_lval to handle method member access.
- Conservative approach to reference counting.
closes: #13309
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Ernie Rael <errael@raelity.com>
|
|
Problem: Vim9: segfault with islocked()
Solution: Check that the lval pointer is not null for objects and
class variables
closes: #13295
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Ernie Rael <errael@raelity.com>
|
|
Problem: No support for writing extended attributes
Solution: Add extended attribute support for linux
It's been a long standing issue, that if you write a file with extended
attributes and backupcopy is set to no, the file will loose the extended
attributes.
So this patch adds support for retrieving the extended attributes and
copying it to the new file. It currently only works on linux, mainly
because I don't know the different APIs for other systems (BSD, MacOSX and
Solaris). On linux, this should be supported since Kernel 2.4 or
something, so this should be pretty safe to use now.
Enable the extended attribute support with normal builds.
I also added it explicitly to the :version output as well as make it
able to check using `:echo has("xattr")`, to have users easily check
that this is available.
In contrast to the similar support for SELINUX and SMACK support (which
also internally uses extended attributes), I have made this a FEAT_XATTR
define, instead of the similar HAVE_XATTR.
Add a test and change CI to include relevant packages so that CI can
test that extended attributes are correctly written.
closes: #306
closes: #13203
Signed-off-by: Christian Brabandt <cb@256bit.org>
|
|
Problem: Vim9 instanceof() fails in a def func
Solution: allow Objects in compile time check
closes: #12907
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
|
|
Problem: Vim9: need instanceof() function
Solution: Implement instanceof() builtin
Implemented in the same form as Python's isinstance because it allows
for checking multiple class types at the same time.
closes: #12867
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: LemonBoy <thatlemon@gmail.com>
|
|
Problem: No support for stable Python 3 ABI
Solution: Support Python 3 stable ABI
Commits:
1) Support Python 3 stable ABI to allow mixed version interoperatbility
Vim currently supports embedding Python for use with plugins, and the
"dynamic" linking option allows the user to specify a locally installed
version of Python by setting `pythonthreedll`. However, one caveat is
that the Python 3 libs are not binary compatible across minor versions,
and mixing versions can potentially be dangerous (e.g. let's say Vim was
linked against the Python 3.10 SDK, but the user sets `pythonthreedll`
to a 3.11 lib). Usually, nothing bad happens, but in theory this could
lead to crashes, memory corruption, and other unpredictable behaviors.
It's also difficult for the user to tell something is wrong because Vim
has no way of reporting what Python 3 version Vim was linked with.
For Vim installed via a package manager, this usually isn't an issue
because all the dependencies would already be figured out. For prebuilt
Vim binaries like MacVim (my motivation for working on this), AppImage,
and Win32 installer this could potentially be an issue as usually a
single binary is distributed. This is more tricky when a new Python
version is released, as there's a chicken-and-egg issue with deciding
what Python version to build against and hard to keep in sync when a new
Python version just drops and we have a mix of users of different Python
versions, and a user just blindly upgrading to a new Python could lead to
bad interactions with Vim.
Python 3 does have a solution for this problem: stable ABI / limited API
(see https://docs.python.org/3/c-api/stable.html). The C SDK limits the
API to a set of functions that are promised to be stable across
versions. This pull request adds an ifdef config that allows us to turn
it on when building Vim. Vim binaries built with this option should be
safe to freely link with any Python 3 libraies without having the
constraint of having to use the same minor version.
Note: Python 2 has no such concept and this doesn't change how Python 2
integration works (not that there is going to be a new version of Python
2 that would cause compatibility issues in the future anyway).
---
Technical details:
======
The stable ABI can be accessed when we compile with the Python 3 limited
API (by defining `Py_LIMITED_API`). The Python 3 code (in `if_python3.c`
and `if_py_both.h`) would now handle this and switch to limited API
mode. Without it set, Vim will still use the full API as before so this
is an opt-in change.
The main difference is that `PyType_Object` is now an opaque struct that
we can't directly create "static types" out of, and we have to create
type objects as "heap types" instead. This is because the struct is not
stable and changes from version to version (e.g. 3.8 added a
`tp_vectorcall` field to it). I had to change all the types to be
allocated on the heap instead with just a pointer to them.
Other functions are also simply missing in limited API, or they are
introduced too late (e.g. `PyUnicode_AsUTF8AndSize` in 3.10) to it that
we need some other ways to do the same thing, so I had to abstract a few
things into macros, and sometimes re-implement functions like
`PyObject_NEW`.
One caveat is that in limited API, `OutputType` (used for replacing
`sys.stdout`) no longer inherits from `PyStdPrinter_Type` which I don't
think has any real issue other than minor differences in how they
convert to a string and missing a couple functions like `mode()` and
`fileno()`.
Also fixed an existing bug where `tp_basicsize` was set incorrectly for
`BufferObject`, `TabListObject, `WinListObject`.
Technically, there could be a small performance drop, there is a little
more indirection with accessing type objects, and some APIs like
`PyUnicode_AsUTF8AndSize` are missing, but in practice I didn't see any
difference, and any well-written Python plugin should try to avoid
excessing callbacks to the `vim` module in Python anyway.
I only tested limited API mode down to Python 3.7, which seemes to
compile and work fine. I haven't tried earlier Python versions.
2) Fix PyIter_Check on older Python vers / type##Ptr unused warning
For PyIter_Check, older versions exposed them as either macros (used in
full API), or a function (for use in limited API). A previous change
exposed PyIter_Check to the dynamic build because Python just moved it
to function-only in 3.10 anyway. Because of that, just make sure we
always grab the function in dynamic builds in earlier versions since
that's what Python eventually did anyway.
3) Move Py_LIMITED_API define to configure script
Can now use --with-python-stable-abi flag to customize what stable ABI
version to target. Can also use an env var to do so as well.
4) Show +python/dyn-stable in :version, and allow has() feature query
Not sure if the "/dyn-stable" suffix would break things, or whether we
should do it another way. Or just don't show it in version and rely on
has() feature checking.
5) Documentation first draft. Still need to implement v:python3_version
6) Fix PyIter_Check build breaks when compiling against Python 3.8
7) Add CI coverage stable ABI on Linux/Windows / make configurable on Windows
This adds configurable options for Windows make files (both MinGW and
MSVC). CI will also now exercise both traditional full API and stable
ABI for Linux and Windows in the matrix for coverage.
Also added a "dynamic" option to Linux matrix as a drive-by change to
make other scripting languages like Ruby / Perl testable under both
static and dynamic builds.
8) Fix inaccuracy in Windows docs
Python's own docs are confusing but you don't actually want to use
`python3.dll` for the dynamic linkage.
9) Add generated autoconf file
10) Add v:python3_version support
This variable indicates the version of Python3 that Vim was built
against (PY_VERSION_HEX), and will be useful to check whether the Python
library you are loading in dynamically actually fits it. When built with
stable ABI, it will be the limited ABI version instead
(`Py_LIMITED_API`), which indicates the minimum version of Python 3 the
user should have, rather than the exact match. When stable ABI is used,
we won't be exposing PY_VERSION_HEX in this var because it just doesn't
seem necessary to do so (the whole point of stable ABI is the promise
that it will work across versions), and I don't want to confuse the user
with too many variables.
Also, cleaned up some documentation, and added help tags.
11) Fix Python 3.7 compat issues
Fix a couple issues when using limited API < 3.8
- Crash on exit: In Python 3.7, if a heap-allocated type is destroyed
before all instances are, it would cause a crash later. This happens
when we destroyed `OptionsType` before calling `Py_Finalize` when
using the limited API. To make it worse, later versions changed the
semantics and now each instance has a strong reference to its own type
and the recommendation has changed to have each instance de-ref its
own type and have its type in GC traversal. To avoid dealing with
these cross-version variations, we just don't free the heap type. They
are static types in non-limited-API anyway and are designed to last
through the entirety of the app, and we also don't restart the Python
runtime and therefore do not need it to have absolutely 0 leaks.
See:
- https://docs.python.org/3/whatsnew/3.8.html#changes-in-the-c-api
- https://docs.python.org/3/whatsnew/3.9.html#changes-in-the-c-api
- PyIter_Check: This function is not provided in limited APIs older than
3.8. Previously I was trying to mock it out using manual
PyType_GetSlot() but it was brittle and also does not actually work
properly for static types (it will generate a Python error). Just
return false. It does mean using limited API < 3.8 is not recommended
as you lose the functionality to handle iterators, but from playing
with plugins I couldn't find it to be an issue.
- Fix loading of PyIter_Check so it will be done when limited API < 3.8.
Otherwise loading a 3.7 Python lib will fail even if limited API was
specified to use it.
12) Make sure to only load `PyUnicode_AsUTF8AndSize` in needed in limited API
We don't use this function unless limited API >= 3.10, but we were
loading it regardless. Usually it's ok in Unix-like systems where Python
just has a single lib that we load from, but in Windows where there is a
separate python3.dll this would not work as the symbol would not have
been exposed in this more limited DLL file. This makes it much clearer
under what condition is this function needed.
closes: #12032
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
|
