Closed bchretien closed 9 years ago
I guess the best strategy will be to reset the branch at each release. If the release process is done correctly, the current content of HEAD will be stored in vX.Y.Z/ and we do not really care to access the old unstable documentation anyway.
So my opinion on this matter would be to reset the branch at each release :)
Then this all comes down to the release cycle, which is a bit undefined for now. I guess we can add a milestone for the 3.0 release, and add what we deem important to it. I'll need some new features for my project "soon", so these will probably be a part of it, and then there will still be the move to Eigen::Ref
which is going to be a major change to the API/ABI.
Yes, of course. For now, you can at least drop the history before the last release. I guess it should still remove quite a lot of data.
Ok, now we have both 2.0
and 3.0
available, and I merged most of the commits. It still takes quite a lot of storage:
* HEAD: 8.3 MiB
* v2.0: 27.2 MiB
* v3.0: 8.3 MiB
I guess re-generating the documentation for RobOptim 2.0 may lower the final size, but for now it should be good enough.
The
gh-pages
branch that contains the automatically-generated documentation is getting bigger and bigger since large chunks of the documentation are modified for any commit. We should probably trim the doc once in a while (or even do it automatically when updating the documentation). Else, users will need to use the--single-branch -b master
clone options, or--depth
for a shallow copy, in order to get the source in an acceptable time when using a slow connection.This can be checked and confirmed with:
With all the branches (including
gh-pages
):With just
master
:Also:
With everyting:
With just
master
: