profile
viewpoint
Mathieu Duponchelle MathieuDuponchelle lyon

hotdoc/hotdoc 37

HotDoc: the tastiest API documentation system

centricular/gstcefsrc 10

A simple gstreamer wrapper around Chromium Embedded Framework

centricular/android_rtsp_example 8

Example application using RTSP on android, with a gradle build system

centricular/rtmp-slate-fallback 3

This is an example on how to fallback to a "slate" stream using compositor.

centricular/cairo 2

meson port of https://cgit.freedesktop.org/cairo/

centricular/freetype2 2

fork of https://git.savannah.gnu.org/cgit/freetype/freetype2.git/ with meson build definitions

hotdoc/test_hotdoc 2

This is an example project using hotdoc, which implements html-aware output testing

centricular/fontconfig 1

fork of https://cgit.freedesktop.org/fontconfig/tree/src with meson build definitions

centricular/gst-rtsp-server 1

RTSP server based on GStreamer

centricular/harfbuzz 1

meson port of http://cgit.freedesktop.org/harfbuzz/

issue commenthotdoc/hotdoc

Why pin wheezy.template version?

hey @doronbehar , the reason is that I don't want to deal with issues that may happen only with a specific subset of dependencies, and would rather users install hotdoc from pip in a virtual env, so that it runs with the exact same versions of those dependencies that it has been tested against.

I am of course completely OK with updating the pins to the latest version of each dependency, provided the tests still pass. See https://github.com/hotdoc/hotdoc/issues/208 for networkx, this one will require some work, probably trivial, to port to the new API.

Distributions can then carry a patch to setup.py if their versions of the deps don't match what setup.py requires, ideally making sure that the tests do pass in that case as well.

doronbehar

comment created time in 2 days

create barnchMathieuDuponchelle/totem-ydl-parser

branch : master

created branch time in 3 days

created repositoryMathieuDuponchelle/totem-ydl-parser

created time in 3 days

issue closedhotdoc/hotdoc

Support redirect directive in yaml front matter

input page:

---
redirect: page_name.markdown || absolute_url
...

The content of this page has moved

closed time in 6 days

MathieuDuponchelle

issue commenthotdoc/hotdoc

Support redirect directive in yaml front matter

we did yes, thanks :)

MathieuDuponchelle

comment created time in 6 days

issue commenthotdoc/hotdoc

gi: Methods "shadowing" information is ignored

yes, shadows is currently not handled, patch welcome :)

thiblahute

comment created time in 6 days

issue commenthotdoc/hotdoc

Don't require full path when linking to other .md files in same directory

Heya @fastlorenzo , I didn't address this until now because I considered it a nice to have feature, but if it is blocking for you I can take a closer look ofc. Can you publish your source somewhere so I can fix it with an actual use case to test with?

tp-m

comment created time in 6 days

issue openedhotdoc/hotdoc

Port to networkx 2.0

networkx has a new major version, porting should be fairly straightforward: https://networkx.github.io/documentation/stable/release/migration_guide_from_1.x_to_2.0.html

created time in 18 days

startedplugorgau/bbb-render

started time in a month

issue closedhotdoc/hotdoc

Something requests user location

We got the following report:

Hey, I wanted to quickly pop in and report that the mesonbuild.com website requests the user's location, this seems like an extremely dickish  thing to do, and I cannot think of any sensible reason to request that permission
This happens even with third-party scripts blocked, so I assume some rogue library snuck in

Meson's web site is fully generated with Hotdoc as installed with Pip. AFAICT we don't do any geo-optimization so it should not need the location info.

closed time in a month

jpakkane

issue commenthotdoc/hotdoc

Something requests user location

There is definitely nothing in hotdoc's javascript code that would ever require a location, this much I can guarantee. The generated documentation relies on a few javascript libraries, which you can find a summary of in a README in the build directory since https://github.com/hotdoc/hotdoc/commit/194f05d30630fc5eda53844281393ac84880c3f9 . While I haven't reviewed the code for that handful of libraries, I would find it weird and indeed disturbing that one of them would ask for a location.

Considering you cannot reproduce and no one has ever reported such a problem, my first guess would be that for some reason you didn't actually connect to mesonbuild.com but to some other similarly named website and I'll close this issue, please don't hesitate to reopen if you figure out how to reproduce.

jpakkane

comment created time in a month

issue closedhotdoc/hotdoc

COPYING file missing from tarballs

And the tarballs on github seem to lag behind the pypi ones: https://pypi.org/project/hotdoc/#history vs. https://github.com/hotdoc/hotdoc/releases

(the snapshots at https://github.com/hotdoc/hotdoc/tags aren't usable because the submodules are missing)

closed time in a month

hadess

push eventhotdoc/hotdoc

Mathieu Duponchelle

commit sha 83d0c70ecb9e0224ec67dcf7c6c5cfee8d049288

Update theme submodule Fixes #206

view details

push time in a month

issue closedhotdoc/hotdoc

Named anchors don't account for floating site-wide header

When you go to a link like this: https://mesonbuild.com/Builtin-options.html#b_vscrt-from_buildtype it puts the anchor position at the top of the page, underneath the site-wide header:

image

This is what it should actually show:

image

The anchor was added manually with <a>:

<a name="b_vscrt-from_buildtype"></a>
The default value of `b_vscrt` is `from_buildtype`. In that case, the following
table is used internally to pick the CRT compiler arguments based on the value
of the `buildtype` option:

closed time in a month

nirbheek

push eventhotdoc/hotdoc_bootstrap_theme

Mathieu Duponchelle

commit sha 67cf9fd3e1b09606a20193995c9160538aadd94c

navbar_offset_scroller: misc improvements + get rid of animated scrolling, feels snappier that way + remove a fixed offset from the scroll destination + support scrolling to a[name] anchors

view details

push time in a month

pull request commentned14/pcpp

preprocessor: preserve line continuations

No problem, didn't mean to be pushy sorry, I hope things sort themselves out on your end eventually :) It's not a blocker for me as replacing the source scanner in gobject-introspection is a large undertaking anyway, and not something I get paid to do either, it is more of a long term project that should have many benefits in terms of maintenance, and a fun side project to hack on in my free time really.

I put up my WIP code at https://gitlab.gnome.org/GNOME/gobject-introspection/-/merge_requests/231 in case you're interested when you find free time again :)

MathieuDuponchelle

comment created time in 2 months

push eventMathieuDuponchelle/pcpp

Mathieu Duponchelle

commit sha 4b35340d9490546b74c3b737555f6d176bc012ed

preprocessor: open included files as utf8

view details

push time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

hey @ned14 , feeling like giving this a look again? :)

MathieuDuponchelle

comment created time in 2 months

push eventhotdoc/hotdoc

Mathieu Duponchelle

commit sha 194f05d30630fc5eda53844281393ac84880c3f9

license_extension: allow theme and extension to advertise licensing

view details

Mathieu Duponchelle

commit sha 59bef853dcba56bdbb75b119fa16c1ce28eca239

Update theme submodule

view details

push time in 2 months

push eventhotdoc/hotdoc_bootstrap_theme

Mathieu Duponchelle

commit sha 7ea9ab5e6fab6412ead100b4c222d5e03866b6da

advertise licensing through theme.json

view details

push time in 2 months

push eventhotdoc/hotdoc

Mathieu Duponchelle

commit sha fc4de1cd343cf2868b41c790f5d57969f58dc038

search_module: fix context id identification eg, in a page with this layout: <div id="foo"> <h2 id="bar"> <p>token</p> </div> Searching for token should link to #bar, not #foo. This was a regression from the C port.

view details

push time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

Is it in your culture such begin of the sentence means order to do something????

Let's say that your phrasing is unusual and leave it at that

kloczek

comment created time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

Please stop focusing on me (I'm not the subject here) and try to do what I've been asking you which is submitting your adaptation to cmock.

What makes you think you can order me around? If you need something done just do it, otherwise be nice and courteous and answer the questions whoever you're demanding something from is asking!

I've not been trying to build hotdoc with standalone cmock because you are not offering such option and for now it will be necessary to modify hotdoc to even try to build that that way.

This sentence makes no sense. You cannot compile against regular cmark, if you had taken the time to look at the PR @thiblahute linked, you would have noticed it adds new API.

Q: it is any API obstacle in way how you are using cmock which would disallow now such build?

see above

kloczek

comment created time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

@kloczek I can't tell if you're a troll or just very good at making people annoyed, in doubt:

You should tell us a bit more about your use case, what distribution are you making a package for? What's the policy of that distribution with respect to bundling? What are the conditions for making exceptions? Do you have any technical difficulties with building hotdoc? Have you checked out the pypi package which has all the sources in tree?

The only question you've half answered is with:

Yes I have dificulties,. One of those difficulties is relateto to be not able use regular cmark. cmark is used by several other packages (hotdoc is not only consumer of that library).

But that's a very vague statement, not even in answer to my question, not being able to use "regular" cmark isn't a blocker for building hotdoc, simply potentially an issue for getting the package accepted.

kloczek

comment created time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

@kloczek if you won't answer any of my questions except in the vaguest of terms, I don't see much of a point in continuing this discussion.

kloczek

comment created time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

@kloczek I understand bundling is considered problematic, but pragmatism has its virtues, we can't force a maintainer to accept a patchset he's not interested in. You should tell us a bit more about your use case, what distribution are you making a package for? What's the policy of that distribution with respect to bundling? What are the conditions for making exceptions? Do you have any technical difficulties with building hotdoc? Have you checked out the pypi package which has all the sources in tree?

kloczek

comment created time in 2 months

issue commenthotdoc/hotdoc

0.11.1: build fails

A few times yes. github ended up cherry-picking those in their own cmark fork, but the cmark maintainer was worried about intrusiveness.

kloczek

comment created time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

From a quick scan, it seems reasonable. But I'll need to find an hour somewhere to properly study it. That can take a while, alas working on pcpp cannot be claimed as work time in any capacity, so it needs to come from when I am not tired, in my free time.

Makes sense, I'm also doing this on my free time fwiw :)

MathieuDuponchelle

comment created time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

loks good to you @ned14 ?

MathieuDuponchelle

comment created time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

I implemented your suggestion, the test suite now passes without modifications. I have also added a simple commit to propagate source / lineno information to the Macro instances, it is trivial so I figured I could keep it in that MR, let me know if you'd rather I make a separate one :)

MathieuDuponchelle

comment created time in 2 months

push eventMathieuDuponchelle/pcpp

Mathieu Duponchelle

commit sha 28dec1a2f0e167fd6e1bad1baa9e8d4cd9e0a858

preprocessor: set source and lineno on created macros

view details

Mathieu Duponchelle

commit sha 9bd36d517ff59b424afb609cb452b8780e4bf005

preprocessor: preserve line continuations Instead of concatenating lines ended with a backslash for further lexing, preserve those by considering them as just another type of whitespace. This causes yielded lex tokens for multiline macros to have an accurate reported line number.

view details

push time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

Your PR as it currently is breaks conformance of output with clang's and GCC's preprocessors.

Yes that's what I suspected right after I posted it, hence my follow-up comment :)

Obviously a great number of people care a lot about whitespace being perfect, and not folding backslashes would be a showstopper for them.

Right, I don't want to break others' use cases of course.

That said, I'm not unsympathetic to your use case.

cool :)

You'd then have to do a bunch of work to ensure that backslash + newline token acts mostly like whitespace, except where it doesn't, and then in the final rendition does collapse into whitespace without affecting collapse of neighbouring whitespace in any way.

That sounds in line with what I had in mind (stripping line continuations at a later time), though my perhaps naive hope was that it wouldn't prove too difficult. I'll poke around some more.

Then you'd add an option to inhibit that final collapse.

For what it's worth, that's not actually an option I'd need, my current approach is to run Preprocessor.parse(), consume all the tokens, then iterate Preprocessor.macros, without calling Preprocessor.write().

I'll be honest, I think that's a lot of work when you have an easier alternative: just apply clang-format to the extracted preprocessed output. That'll reformat your macros into the shape you had above i.e. backslashes after lines. It won't be one-one with the source, but it'll be probably even better from a docs generation perspective because it'll make bad original formatting pretty. I'll leave the choice up to you.

That's an interesting point, but I think the correct approach at the gobject-introspection level is to serialize (as close as possible to) the original source, leaving formatting decisions up to tools further downstream. I'll think about it if the alternative proves too annoying / intrusive to implement, thanks :)

MathieuDuponchelle

comment created time in 2 months

pull request commentned14/pcpp

preprocessor: preserve line continuations

Actually, I don't know if the changes in the reference output are legitimate, or if line continuations should be stripped out at a later stage? Let me know if I should do that, for context I'm currently investigating whether pcpp could be used as a replacement for the macro / comment scanner in gobject-introspection (https://gitlab.gnome.org/GNOME/gobject-introspection/-/issues/346), and one feature that I consider a requirement is the ability to serialize object macros / function macros verbatim in the output XML, eg given:

#define foo() { \
  bar(); \
}

I would like the output xml to look similar to:

  <function-macro name="foo">{
  bar(); \
}</function-macro>
MathieuDuponchelle

comment created time in 2 months

PR opened ned14/pcpp

preprocessor: preserve line continuations

Instead of concatenating lines ended with a backslash for further lexing, preserve those by considering them as just another type of whitespace. This causes yielded lex tokens for multiline macros to have an accurate reported line number.

The changes in the reference output for tests are legitimate.

+36 -23

0 comment

4 changed files

pr created time in 2 months

create barnchMathieuDuponchelle/pcpp

branch : preserve-line-continuations

created branch time in 2 months

fork MathieuDuponchelle/pcpp

A C99 preprocessor written in pure Python

fork in 2 months

push eventhotdoc/hotdoc

Thibault Saunier

commit sha 28ad9aa59daa49d6b3fd1bc6b144df52af39475b

gtk_doc: Support annotation lines ended with colons

view details

push time in 3 months

push eventhotdoc/hotdoc

Thibault Saunier

commit sha 018ff4a0a3562c9e29ab8f99c6914beca8d09c1c

database: Allow aliases to have the same name as a real symbol GstAppSrc is appsrcy

view details

push time in 3 months

PR merged hotdoc/hotdoc

database: Allow aliases to have the same name as a real symbol

GstAppSrc is appsrcy

+18 -16

0 comment

2 changed files

thiblahute

pr closed time in 3 months

pull request commentszatmary/libcaption

caption_frame_decode: rework API

ping @szatmary :)

MathieuDuponchelle

comment created time in 3 months

push eventhotdoc/hotdoc

Thibault Saunier

commit sha 95aa79758f64281de3f0e6d2efde45108d8cd2ea

gst: Use plugin name to reference plugins Instead of the filename

view details

push time in 3 months

PR merged hotdoc/hotdoc

gst: Use plugin name to reference plugins

Instead of the filename

+3 -2

0 comment

1 changed file

thiblahute

pr closed time in 3 months

push eventhotdoc/hotdoc

Thibault Saunier

commit sha 0171b68bcb540d2cd74d833f652bf95ebdc4d74f

gst: Move other class types to their own pages And properly use GIClassSymbol for other classed types as they actually are GObject and we need the surounding logic there.

view details

push time in 3 months

PR merged hotdoc/hotdoc

gst: Move other class types to their own pages

And properly use GIClassSymbol for other classed types as they actually are GObject and we need the surounding logic there.

+26 -13

0 comment

2 changed files

thiblahute

pr closed time in 3 months

more