Shorter and easier to read while debugging

currently, we can't look correctly for virtual functions/class methods
as they are manually impl...

this approaches uses analysis so we are sure the item do exists

ObjectExt is in glib's prelude

gstreamer has some ignored enum members which ends up in the docs links
and we need a way to ign...

instead of passing the current type name, we use the TypeId for that
and now we render Self::som...

* fixup: rsplit+next

* fixup: Add back missing () on unresolved function symbols

* fixup: ...

this also fixes records search

the helper checks whether a function can be used as doc link.
and some stuff that are duplicated...

This was overriding the `type_id` for enums and bitfields so that they
always resolve to the las...

because gi-docgen uses @ in it's format, it avoids breaking things before they are replaced by g...

instead of always returning the symbol name formatted we return an Option
in case we want to cha...

like the functions::Info's doc_link but for specific types
the helpers are here to ensure both l...

as we don't mark explicitly as ignored on GIR side

with this approach we can only omit links when it makes sense instead of always having one

a parser and replacer for gi-docgen special links to rust ones

it's very common to have GskX or PangoX
with this change both will be replaced with their equiva...

the reasoning behind is we want to be able to replace `GtkButton` with a link to Button
instead ...

This will be helpful for later so we can use analysis to find a type by it's rust name
or it's k...

flags: generate doc aliases

enums,flags: Do not analyze imports if the type is not generated

Otherwise unused imports show up in the enums/flags files.

Fixes: 68591f1 ("enums,flags: Always...

enums,flags: Always analyze manual types

codegen/doc: Make extension trait docs link back to the type

Otherwise some enum members are not found

docs: filter out final types from implements trait list

The name is already there, but it isn't an intradoc-link yet.

Because such traits don't exists for those types

Docs generation is going to look at analysis data to know what variants
and functions are availa...

When the surrounding type is ignored or manually implemented the
underlying functions are never ...

[refactor] Simplify handling of trait bounds and aliases

Most of the callers were not reading the return value leading to
potentially missed failed type ...

Regardless of it being a generic type with the first type parameter or
not.

This member was always set to `false`, and by extension `fn
get_base_alias` always returns `None`.

This simplifies the code (gir and generated results) a fair bit by
placing the trait bound direc...

codgen: generate impl T if there are builder properties as well

Fixes https://github.com/gtk-rs/gtk4-rs/issues/405

codgen: generate a builder method for objects with a Builder

fixes #1070

docs: use doc_ignore_parameters for global functions as well

docs: fix a missing param due to not rebasing #1151

docs: ignore not useful function parameters

docs: generate docs for global functions

codegen: use C value instead of hardcoding it

this allows defining a list of parameters to be ignored when the docs are generated
other than t...

This changes few things in the generated docs of a function
- ignores a bunch of function parame...

codegen/doc: Drop `[Deprecated]` text in favour of Rust annotations

docs: generate docs for builder properties

codegen/object: Add missing deprecation attribute to builders

All deprecated objects, traits and records already receive a
`#[deprecated = "Since XXX"]` annot...

Rename `do_main` to `main`

This app was probably written before Rust 1.26 introduced main who could return Result
https://g...

codegen: default to gtk-rs-core for sys crates

Stop dereferencing followed by reborrowing in `match` and `if let`

codegen/sys: Use pointer formatting literal to print address of self

Prints the same, but without a cast.

Use try_from_glib shortcut

Update CI

Fix some clippy/compiler warnings

funct async: use codegen name for _future funct

... otherwise we can't take advantage of config and auto-renaming
on non-future function.