Some important macros are missing from geany-gtkdoc.h, mainly GEANY_WORDCHARS.
Extend the generation script to add those to geany-gtkdoc.h
Note: g-ir-scanner ignores most of the macros, except those that are plain defines for string literals or numbers/characters, so only a few are usable for GIR.
Note also: Due to a bug in g-ir-scanner, GEANY_ABI_VERSION is parsed incorrectly (simply without the GTK version shift) but the ABI is of little relevance for GIR consumers. You can view, comment on, or merge this pull request online at:
https://github.com/geany/geany/pull/1796
-- Commit Summary --
* gtkdoc: include macros in geany-gtkdoc.h
-- File Changes --
M scripts/gen-api-gtkdoc.py (26)
-- Patch Links --
https://github.com/geany/geany/pull/1796.patch https://github.com/geany/geany/pull/1796.diff
b4n requested changes on this pull request.
Is `GEANY_WORDCHARS` actually useful to anyone? I think it's pretty much useless in most situations, and wrong in many other (e.g. it's *not* a good universal way to recognize a "word", not even for programming languages). And it's fairly easy to copy over, or even generate using I don't know, `'_' + string.lowercase + string.uppercase + string.digits` in Python, and probably something similar in another language.
Also, most if not all the other macros are irrelevant to non-C, like `foreach_document`, `documents` and alike. The only potentially interesting things would be `TAG_*`, but I'm not even sure we really think it's part of the API (OK, it is, but I'm not sure we did know), and no plugin is using it ATM AFAICT.
`GEANY_API_VERSION` is also probably uninteresting as I guess the proxy is more relevant here.
Anyway, the code here seem mostly good to me, but I'm not very sure it's a good or useful idea. And if we do want macros, we should probably mark a lot of the existing ones `@girskip` in any case.
Note also: Due to a bug in g-ir-scanner, GEANY_ABI_VERSION is parsed
incorrectly (simply without the GTK version shift) but the ABI is of little relevance for GIR consumers.
FWIW `GEANY_ABI_SHIFT` is not part of the API, and thus is not in the gtkdoc header, no surprise g-ir-scanner has a hard time :)
+class DoxyDefine(DoxyElement):
+ @staticmethod + def from_memberdef(xml): + name = xml.find("name").text + v = xml.find("initializer") + if (v is not None): + proc = DoxygenProcess() + text = proc.process_element(v) + params = ",".join(p.text for p in xml.xpath(".//param/defname")) + if (params != ""): + params = "(%s)" % params + d = "#define %s%s %s" % (name, params, text) + else: + d = "#define %s" % name + + e = DoxyEnum(name, d)
Shouldn't this be `DocyDefine(name, d)`? (not that it actually matters much, but still)
sagarchalise commented on this pull request.
+class DoxyDefine(DoxyElement):
+ @staticmethod + def from_memberdef(xml): + name = xml.find("name").text + v = xml.find("initializer") + if (v is not None): + proc = DoxygenProcess() + text = proc.process_element(v) + params = ",".join(p.text for p in xml.xpath(".//param/defname")) + if (params != ""): + params = "(%s)" % params + d = "#define %s%s %s" % (name, params, text) + else: + d = "#define %s" % name + + e = DoxyEnum(name, d)
I was the one with the idea of providing this in GI. Basically, it is provided in plugin documentation and I thought it would be consistent to provide it with geany gi. Also, I use it in a project and you are correct that I can easily include it as well as customize it.
https://github.com/sagarchalise/geany-pyjedi/blob/peasy/peasyjedi.py
github-comments@lists.geany.org