Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Apr 15 22:32
    stsewd closed #8111
  • Apr 15 22:30
    stsewd review_requested #8084
  • Apr 15 22:28
    stsewd synchronize #8084
  • Apr 15 21:54
    stsewd closed #8110
  • Apr 15 21:48
    stsewd auto_merge_enabled #8110
  • Apr 15 21:47
    stsewd synchronize #8110
  • Apr 15 21:39
    stsewd review_requested #8111
  • Apr 15 21:38
    stsewd opened #8111
  • Apr 15 19:45
    agjohnson synchronize #8109
  • Apr 15 18:25
    stsewd review_requested #8110
  • Apr 15 18:25
    stsewd opened #8110
  • Apr 15 17:53
    stsewd review_requested #7808
  • Apr 15 17:35
    agjohnson opened #8109
  • Apr 15 17:12
    stsewd synchronize #8103
  • Apr 15 14:27
    stsewd closed #8104
  • Apr 15 10:11
    astrojuanlu closed #8082
  • Apr 15 00:58
    cocobennett synchronize #8082
  • Apr 14 23:47
    cocobennett synchronize #8082
  • Apr 14 23:44
    cocobennett synchronize #8082
  • Apr 14 22:28
    stsewd review_requested #8107
Aaron Carlisle
@Blendify
Downloads of 1.7 are very little so dropping that wont matter much
Aaron Carlisle
@Blendify
I think dropping 1.6 and 1.7 would make sense in this case.
Also since old projects are pinned to an old version of sphinx dropping support for newer sphinx versions wont affect the old projects
@agjohnson regardless of what we decide we need to fix some things for sphinx4. I made a patch here: https://github.com/readthedocs/sphinx_rtd_theme/pull/1123/files
Anthony
@agjohnson
well, old sphinx version doesn't necessarily mean old theme version
we have some defaults if the project doesn't use a requirements file/etc, but it's absolutely possible to combine sphinx 1.6 and theme 0.5.2 even
we have no data to determine the fallout from this unfortunately
Aaron Carlisle
@Blendify
My opinion if a user writes their own requirements file they are responsible for making sure their config doesn't break with new versions
Anthony
@agjohnson
heh, well we get user and customer complaints when sites change or builds break, so it's an area we try to be a bit on the cautious side
though, even with proper deprecation notices and semver, we can't assume that most users will pick up on deprecations even
Aaron Carlisle
@Blendify
understandable, especially with the commercial side of rtd
Anthony
@agjohnson
if we had been collecting data on sphinx + sphinx theme version usage, this would be a lot easier
we did add this to our roadmap, but doesn't help us immediately
the best outcome for any of these scenarios would be that the build should fail entirely
Aaron Carlisle
@Blendify
Dropping sphinx 1.6 and 1.7 could be a test also to judge the fallout of a bigger breaking change
Anthony
@agjohnson
if we hit a version mismatch issue, or some other deprecation, etc
i've lost track, what requires us to drop 1.6 and 1.7?
could we instead float along with 1.6 and 1.7 until we drop all of 1.x and go 2+?
Aaron Carlisle
@Blendify
Nothing particular it just breaks up the number of affected users
Anthony
@agjohnson
aye
Aaron Carlisle
@Blendify
dropping 1.6 and 1.7 would only affect about half the installs (in theory) if we were to just drop 1.x
Both 1.6 and 1.8 have about 15k downloads per month so dropping 1.6 and 1.7 would drop ~15k while dropping 1.x would drop ~30k
its hard to say though because this is just sphinx and doesn't link to rtd or which theme the users use.
Anthony
@agjohnson
yeah absolutely, i regret not tracking more build metadata here sooner
Aaron Carlisle
@Blendify
I am just worried of the sphinx2 jump because that drops html4 and python2 I don't want to make that a huge burden when we don't know what the fallout would be for even dropping 1.6x
Anthony
@agjohnson
we can do sphinx 2 + html4 -- html4 is dropped in ... 3.0?
Aaron Carlisle
@Blendify
and drop 1.x in v2?
Anthony
@agjohnson
yeah, i think so?
not sure which direction i'm leaning towards at the moment
Aaron Carlisle
@Blendify
Yeah could work, but doesn't really solve anything except getting v1 released sooner
I was kinda partial to matching the major theme version with the major sphinx version
So if users want to use theme v2 they know they need sphinx v2
Anthony
@agjohnson
ah yeah, i don't think we'd be able to keep that up for too long
i'm sort of leaning towards keeping sphinx deprecations out of the picture until we have a reason to -- like dropping html4 writer support (with the bootstrap theme release?)
that is, figure out what we want to fix before closing the development line that include wyrm styles, and then do a big step up
Aaron Carlisle
@Blendify
Well technically we dont even use the html4 doctype so we currently kinda depend on sphinx2
It seems the theme never really used the html4 doctype :/
Aaron Carlisle
@Blendify
So technically we could use bootstrap with 1.6 or anyversion. Its just easier to support sphinx to prevent the use of if statements everywhere. https://github.com/readthedocs/sphinx_rtd_theme/blob/master/sphinx_rtd_theme/__init__.py#L36 is kinda becoming a mess
Anthony
@agjohnson
yeah for the theme, html4 support was more like is more like "not html5"
Aaron Carlisle
@Blendify
lol
Anthony
@agjohnson
i was just considering if it made sense to support both wyrm/bootstrap for a release or two, and make the theme selectable with an option or something
i can't decide if that is better or worse for maintenance though
Aaron Carlisle
@Blendify
I dont think that really solves anything, I think could make it worse for maintenance. RIpping off the bandaid would be better.
Dalton Smith
@Daltz333
@stsewd got hoverxref working for mathjax3
image.png
It wasn't horribly difficult to figure out, although MathJax API docs are pretty garbage. I wanted to run it async but that didn't turn out very good with the hover tooltip, so I opted out of that. You might notice that it doesn't update a specific element anymore. That's because MathJax3 caches current vs changed math, so it's not necessary to update a specific element. Additionally, their promise().then() API which is for updating a specific element is only when you want to update one specific thing that changed, versus all things that changed... additionally, I couldn't figure it out.
Manuel Kaufmann
@humitos
@Daltz333 thanks for your PR! I took a quick look and it's great! I'll do a full review soon hopefully