Mercurial > vim
view runtime/doc/textprop.txt @ 32936:c517845bd10e v9.0.1776
patch 9.0.1776: No support for stable Python 3 ABI
Commit: https://github.com/vim/vim/commit/c13b3d1350b60b94fe87f0761ea31c0e7fb6ebf3
Author: Yee Cheng Chin <ychin.git@gmail.com>
Date: Sun Aug 20 21:18:38 2023 +0200
patch 9.0.1776: No support for stable Python 3 ABI
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>
author | Christian Brabandt <cb@256bit.org> |
---|---|
date | Sun, 20 Aug 2023 21:30:04 +0200 |
parents | 98db7b3b4827 |
children | c7a332a34fe7 |
line wrap: on
line source
*textprop.txt* For Vim version 9.0. Last change: 2023 Apr 23 VIM REFERENCE MANUAL by Bram Moolenaar Displaying text with properties attached. *textprop* *text-properties* 1. Introduction |text-prop-intro| 2. Functions |text-prop-functions| 3. When text changes |text-prop-changes| {not able to use text properties when the |+textprop| feature was disabled at compile time} ============================================================================== 1. Introduction *text-prop-intro* Text properties can be attached to text in a buffer. They will move with the text: If lines are deleted or inserted the properties move with the text they are attached to. Also when inserting/deleting text in the line before the text property. And when inserting/deleting text inside the text property, it will increase/decrease in size. The main use for text properties is to highlight text. This can be seen as a replacement for syntax highlighting. Instead of defining patterns to match the text, the highlighting is set by a script, possibly using the output of an external parser. This only needs to be done once, not every time when redrawing the screen, thus can be much faster, after the initial cost of attaching the text properties. Text properties can also be used for other purposes to identify text. For example, add a text property on a function name, so that a search can be defined to jump to the next/previous function. A text property is attached at a specific line and column, and has a specified length. The property can span multiple lines. A text property has these fields: "id" a number to be used as desired "type" the name of a property type Property Types ~ *E971* A text property normally has the name of a property type, which defines how to highlight the text. The property type can have these entries: "highlight" name of the highlight group to use "combine" when omitted or TRUE the text property highlighting is combined with any syntax highlighting; when FALSE the text property highlighting replaces the syntax highlighting "priority" when properties overlap, the one with the highest priority will be used. "start_incl" when TRUE inserts at the start position will be included in the text property "end_incl" when TRUE inserts at the end position will be included in the text property Example ~ Suppose line 11 in a buffer has this text (excluding the indent): The number 123 is smaller than 4567. To highlight the numbers in this text: > call prop_type_add('number', {'highlight': 'Constant'}) call prop_add(11, 12, {'length': 3, 'type': 'number'}) call prop_add(11, 32, {'length': 4, 'type': 'number'}) Try inserting or deleting lines above the text, you will see that the text properties stick to the text, thus the line number is adjusted as needed. Setting "start_incl" and "end_incl" is useful when white space surrounds the text, e.g. for a function name. Using false is useful when the text starts and/or ends with a specific character, such as the quote surrounding a string. func FuncName(arg) ~ ^^^^^^^^ property with start_incl and end_incl set var = "text"; ~ ^^^^^^ property with start_incl and end_incl not set Nevertheless, when text is inserted or deleted the text may need to be parsed and the text properties updated. But this can be done asynchronously. Internal error *E967* If you see E967, please report the bug. You can do this at Github: https://github.com/vim/vim/issues/new ============================================================================== 2. Functions *text-prop-functions* Manipulating text property types: prop_type_add({name}, {props}) define a new property type prop_type_change({name}, {props}) change an existing property type prop_type_delete({name} [, {props}]) delete a property type prop_type_get({name} [, {props}]) get property type values prop_type_list([{props}]) get list of property types Manipulating text properties: prop_add({lnum}, {col}, {props}) add a text property prop_add_list({props}, [{item}, ...]) add a text property at multiple positions. prop_clear({lnum} [, {lnum-end} [, {bufnr}]]) remove all text properties prop_find({props} [, {direction}]) search for a text property prop_list({lnum} [, {props}]) text properties in {lnum} prop_remove({props} [, {lnum} [, {lnum-end}]]) remove a text property *prop_add()* *E965* prop_add({lnum}, {col}, {props}) Attach a text property at position {lnum}, {col}. {col} is counted in bytes, use one for the first column. If {lnum} is invalid an error is given. *E966* If {col} is invalid an error is given. *E964* {props} is a dictionary with these fields: type name of the text property type length length of text in bytes, can only be used for a property that does not continue in another line; can be zero end_lnum line number for the end of text (inclusive) end_col column just after the text; not used when "length" is present; when {col} and "end_col" are equal, and "end_lnum" is omitted or equal to {lnum}, this is a zero-width text property bufnr buffer to add the property to; when omitted the current buffer is used id user defined ID for the property; must be a number, should be positive; when using "text" then "id" must not be present and will be set automatically to a negative number; otherwise zero is used *E1305* text text to be displayed before {col}, or above/below the line if {col} is zero; prepend and/or append spaces for padding with highlighting; cannot be used with "length", "end_lnum" and "end_col" See |virtual-text| for more information. *E1294* text_align when "text" is present and {col} is zero; specifies where to display the text: after after the end of the line right right aligned in the window (unless the text wraps to the next screen line) below in the next screen line above just above the line When omitted "after" is used. Only one "right" property can fit in each line, if there are two or more these will go in a separate line (still right aligned). text_padding_left *E1296* used when "text" is present and {col} is zero; padding between the end of the text line (leftmost column for "above" and "below") and the virtual text, not highlighted text_wrap when "text" is present and {col} is zero, specifies what happens if the text doesn't fit: wrap wrap the text to the next line truncate truncate the text to make it fit When omitted "truncate" is used. Note that this applies to the individual text property, the 'wrap' option sets the overall behavior All fields except "type" are optional. It is an error when both "length" and "end_lnum" or "end_col" are given. Either use "length" or "end_col" for a property within one line, or use "end_lnum" and "end_col" for a property that spans more than one line. When neither "length" nor "end_col" are given the property will be zero-width. That means it will move with the text, as a kind of mark. One character will be highlighted, if the type specifies highlighting. The property can end exactly at the last character of the text, or just after it. In the last case, if text is appended to the line, the text property size will increase, also when the property type does not have "end_incl" set. "type" will first be looked up in the buffer the property is added to. When not found, the global property types are used. If not found an error is given. *virtual-text* When "text" is used and the column is non-zero then this text will be displayed at the specified start location of the text property. The text of the buffer line will be shifted to make room. This is called "virtual text". When the column is zero the virtual text will appear above, after or below the buffer text. The "text_align" and "text_wrap" arguments determine how it is displayed. To separate the virtual text from the buffer text prepend and/or append spaces to the "text" field or use the "text_padding_left" value. Make sure to use a highlight that makes clear to the user that this is virtual text, otherwise it will be very confusing that the text cannot be edited. When using "above" you need to make clear this text belongs to the text line below it, when using "below" you need to make sure it belongs to the text line above it. The text will be displayed but it is not part of the actual buffer line, the cursor cannot be placed on it. A mouse click in the text will move the cursor to the first character after the text, or the last character of the line. Any Tab and other control character in the text will be changed to a space (Rationale: otherwise the size of the text is difficult to compute). A negative "id" will be chosen and is returned. Before text properties with text were supported it was possible to use a negative "id", even though this was very rare. Now that negative "id"s are reserved for text properties with text an error is given when using a negative "id". When a text property with text already exists using a negative "id" results in *E1293* . If a negative "id" was used and later a text property with text is added results in *E1339* . Can also be used as a |method|: > GetLnum()->prop_add(col, props) < *prop_add_list()* prop_add_list({props}, [{item}, ...]) Similar to prop_add(), but attaches a text property at multiple positions in a buffer. {props} is a dictionary with these fields: bufnr buffer to add the property to; when omitted the current buffer is used id user defined ID for the property; must be a number; when omitted zero is used type name of the text property type All fields except "type" are optional. The second argument is a List of items, where each {item} is a list that specifies the starting and ending position of the text: [{lnum}, {col}, {end-lnum}, {end-col}] or: [{lnum}, {col}, {end-lnum}, {end-col}, {id}] The first two items {lnum} and {col} specify the starting position of the text where the property will be attached. The next two items {end-lnum} and {end-col} specify the position just after the text. An optional fifth item {id} can be used to give a different ID to a property. When omitted the ID from {props} is used, falling back to zero if none are present. It is not possible to add a text property with a "text" field here. Example: > call prop_add_list(#{type: 'MyProp', id: 2}, \ [[1, 4, 1, 7], \ [1, 15, 1, 20], \ [2, 30, 3, 30]] < Can also be used as a |method|: > GetProp()->prop_add_list([[1, 1, 1, 2], [1, 4, 1, 8]]) prop_clear({lnum} [, {lnum-end} [, {props}]]) *prop_clear()* Remove all text properties from line {lnum}. When {lnum-end} is given, remove all text properties from line {lnum} to {lnum-end} (inclusive). When {props} contains a "bufnr" item use this buffer, otherwise use the current buffer. Can also be used as a |method|: > GetLnum()->prop_clear() < *prop_find()* prop_find({props} [, {direction}]) Search for a text property as specified with {props}: id property with this ID type property with this type name both "id" and "type" must both match bufnr buffer to search in; when present a start position with "lnum" and "col" must be given; when omitted the current buffer is used lnum start in this line (when omitted start at the cursor) col start at this column (when omitted and "lnum" is given: use column 1, otherwise start at the cursor) skipstart do not look for a match at the start position A property matches when either "id" or "type" matches. {direction} can be "f" for forward and "b" for backward. When omitted forward search is performed. If a match is found then a Dict is returned with the entries as with prop_list(), and additionally an "lnum" entry. If no match is found then an empty Dict is returned. prop_list({lnum} [, {props}]) *prop_list()* Returns a List with all the text properties in line {lnum}. The following optional items are supported in {props}: bufnr use this buffer instead of the current buffer end_lnum return text properties in all the lines between {lnum} and {end_lnum} (inclusive). A negative value is used as an offset from the last buffer line; -1 refers to the last buffer line. types List of property type names. Return only text properties that match one of the type names. ids List of property identifiers. Return only text properties with one of these identifiers. The properties are ordered by starting column and priority. Each property is a Dict with these entries: lnum starting line number. Present only when returning text properties between {lnum} and {end_lnum}. col starting column length length in bytes, one more if line break is included id property ID text text to be displayed before {col}. Only present for |virtual-text| properties. text_align alignment property of |virtual-text|. text_wrap specifies whether |virtual-text| is wrapped. type name of the property type, omitted if the type was deleted type_bufnr buffer number for which this type was defined; 0 if the type is global start when TRUE property starts in this line end when TRUE property ends in this line When "start" is zero the property started in a previous line, the current one is a continuation. When "end" is zero the property continues in the next line. The line break after this line is included. Returns an empty list on error. Examples: " get text properties placed in line 5 echo prop_list(5) " get text properties placed in line 20 in buffer 4 echo prop_list(20, {'bufnr': 4}) " get all the text properties between line 1 and 20 echo prop_list(1, {'end_lnum': 20}) " get all the text properties of type 'myprop' echo prop_list(1, {'types': ['myprop'], \ 'end_lnum': -1}) " get all the text properties of type 'prop1' or 'prop2' echo prop_list(1, {'types': ['prop1', 'prop2'], \ 'end_lnum': -1}) " get all the text properties with ID 8 echo prop_list(1, {'ids': [8], 'end_lnum': line('$')}) " get all the text properties with ID 10 and 20 echo prop_list(1, {'ids': [10, 20], 'end_lnum': -1}) " get text properties with type 'myprop' and ID 100 " in buffer 4. echo prop_list(1, {'bufnr': 4, 'types': ['myprop'], \ 'ids': [100], 'end_lnum': -1}) Can also be used as a |method|: > GetLnum()->prop_list() < *prop_remove()* *E968* *E860* prop_remove({props} [, {lnum} [, {lnum-end}]]) Remove a matching text property from line {lnum}. When {lnum-end} is given, remove matching text properties from line {lnum} to {lnum-end} (inclusive). When {lnum} is omitted remove matching text properties from all lines (this requires going over all lines, thus will be a bit slow for a buffer with many lines). {props} is a dictionary with these fields: id remove text properties with this ID type remove text properties with this type name types remove text properties with type names in this List both "id" and "type"/"types" must both match bufnr use this buffer instead of the current one all when TRUE remove all matching text properties, not just the first one Only one of "type" and "types" may be supplied. *E1295* A property matches when either "id" or one of the supplied types matches. If buffer "bufnr" does not exist you get an error message. If buffer "bufnr" is not loaded then nothing happens. Returns the number of properties that were removed. Can also be used as a |method|: > GetProps()->prop_remove() prop_type_add({name}, {props}) *prop_type_add()* *E969* *E970* Add a text property type {name}. If a property type with this name already exists an error is given. Nothing is returned. {props} is a dictionary with these optional fields: bufnr define the property only for this buffer; this avoids name collisions and automatically clears the property types when the buffer is deleted. highlight name of highlight group to use priority when a character has multiple text properties the one with the highest priority will be used; negative values can be used, the default priority is zero combine when omitted or TRUE combine the highlight with any syntax highlight; when FALSE syntax highlight will not be used override when TRUE the highlight overrides any other, including 'cursorline' and Visual start_incl when TRUE inserts at the start position will be included in the text property end_incl when TRUE inserts at the end position will be included in the text property Can also be used as a |method|: > GetPropName()->prop_type_add(props) prop_type_change({name}, {props}) *prop_type_change()* Change properties of an existing text property type. If a property with this name does not exist an error is given. The {props} argument is just like |prop_type_add()|. Can also be used as a |method|: > GetPropName()->prop_type_change(props) prop_type_delete({name} [, {props}]) *prop_type_delete()* Remove the text property type {name}. When text properties using the type {name} are still in place, they will not have an effect and can no longer be removed by name. {props} can contain a "bufnr" item. When it is given, delete a property type from this buffer instead of from the global property types. When text property type {name} is not found there is no error. Can also be used as a |method|: > GetPropName()->prop_type_delete() prop_type_get({name} [, {props}]) *prop_type_get()* Returns the properties of property type {name}. This is a dictionary with the same fields as was given to prop_type_add(). When the property type {name} does not exist, an empty dictionary is returned. {props} can contain a "bufnr" item. When it is given, use this buffer instead of the global property types. Can also be used as a |method|: > GetPropName()->prop_type_get() prop_type_list([{props}]) *prop_type_list()* Returns a list with all property type names. {props} can contain a "bufnr" item. When it is given, use this buffer instead of the global property types. ============================================================================== 3. When text changes *text-prop-changes* Vim will do its best to keep the text properties on the text where it was attached. When inserting or deleting text the properties after the change will move accordingly. When text is deleted and a text property no longer includes any text, it is deleted. However, a text property that was defined as zero-width will remain, unless the whole line is deleted. *E275* When a buffer is unloaded, all the text properties are gone. There is no way to store the properties in a file. You can only re-create them. When a buffer is hidden the text is preserved and so are the text properties. It is not possible to add text properties to an unloaded buffer. When using replace mode, the text properties stay on the same character positions, even though the characters themselves change. To update text properties after the text was changed, install a callback with `listener_add()`. E.g, if your plugin does spell checking, you can have the callback update spelling mistakes in the changed text. Vim will move the properties below the changed text, so that they still highlight the same text, thus you don't need to update these. Text property columns are not updated or copied: ~ - When setting the line with |setline()| or through an interface, such as Lua, Tcl or Python. Vim does not know what text got inserted or deleted. - With a command like `:move`, which takes a line of text out of context. vim:tw=78:ts=8:noet:ft=help:norl: