#readthedocs

hi all, is there a way to use the RTD api to verify a github repo has configured project on RTD?

agj: Hey, do you know the reason behind removing the django-filters?

i think we were only using it for one page and it wasn't django 1.11 compatible?

or something like that

or py3 compatiable

we weren't using it is the main point :)

Damn, I wanted to write a bug report for my build issues of yesterday but the logs changed (!). No longer mentioning a memory issue but just a build timeout on `cat conf.py`...

this sounds like your pip (i'm assuming actually conda?) install step is taking more than 900 seconds in total

Akien: ^

I don't know, we don't have any specific requirements and on normal builds it takes only a few seconds

But right now I'm hitting another issue while trying to reproduce the error I had yesterday: https://readthedocs.org/projects/godot/builds/6...

I'm currently doing another build to see if it's reproducible and I'll file an issue

ah yes, godot!

i've almost played around with this to play with making micro games

too many hobbies at the moment, i suppose

Akien: so we could do a better job of catching this, it's actually failing on the sphinx-build step

due to a timeout

your project timeout looks like 30m currently

Yeah our docs is huge and sphinx is slow, so we might need a higher timeout I guess

wow! i suspected some of that was pip/conda install, but it's all sphinx, huh?

any clue what takes so long?

I guess only the fact that we have a high number of pages (especially with the API reference, ~500 pages or so): https://github.com/godotengine/godot-docs/

But it could also be that we don't use sphinx efficiently, or stress it out too much. In the API reference we have tons of hyperlinks generated for each method/property, that might have an impact on index generation I guess?

agj: Its django 1.11 compatible. Its a good package for filtering in DRF

i don't remember why we removed

i'd avoid adding it back in either way

Akien: it could be just a large document set, yeah

especially if you aren't doing anything with notebooks or such

agj: this issue actually https://github.com/rtfd/readthedocs.org/issues/...

yup, makes more sense now

Akien: we can probably solve this with more build resources, but if your first pass is timing out at 1800s, that means your project needs ~3000s per build, and there are 2-4 builds :x

could you time a fresh build locally? i'm curious how much time is required normally for this

agj: I could not understand the reason to remove and not add it actually!

agj: we're about to release 3.0 so the 2.1 branch will no longer be updated much, if at all. And if we could get stable symlinked to 3.0 instead of building it from the same branch anew, we could spare one build more and have only 3.0/stable + latest

But yeah it's slow. Local builds are quite slow too, I'll have a look in the evening to see how long it takes for me on an i5

In the meantime I could reproduce today's new build error twice, so I'll file a bug report: https://framabin.org/?37d3112f7a8abcae#pZ4WtZh1...

Oh no, it's actually a different issue/stacktrace...

Well, I'll make a "here are my problems" issue then, can't really narrow it down to a very specific and reproducible bug if builds are not reproducible

anyone know how to document a parent class in sphinx? e.g .. py:class:: Foo(...) -- somehow indicate Foo extends "Bar"?

hi all,

is there a way to get the id of the latest build of a project from the api?

I'm hoping to automatically verify the build output