Tesseract clone performance

[haudren]

I was recently updating my local copy of Tesseract and noticed a simple fetch took several minutes ( on a decent but far from amazing network connection). After a little bit of prodding, I noticed the following:

• A regular clone from scratch, e.g. git clone https://github.com/ros-industial-consortium/tesseract clocks in at almost 1.5GiB and 379062 objects which can take hours to download on some unstable networks.
• A smaller clone, git clone https://github.com/ros-industial-consortium/tesseract --single-branch that only retrieves master give me only 27777 objects for a total of 79.99 MiB which seems perfectly reasonable.

I believe the issue is linked to the new (?) gh_pages branch that seems to contain a lot of updates and is a pretty big performance drag.

Would it be possible to host the documentation in some other way? I will try to remember to "ignore" that branch in the meantime, but it makes my git flow a bit less agile 😅

[gavanderhoorn]

I've noticed this as well.

I'm surprised GH hasn't complained yet about the size of the repository.

---

Edit: it's likely the rendered doxygen -- and the many changes to that content -- which have resulted in the increase.

[Levi-Armstrong]

Thank you for the information. We have plans to move the documentation to its own repository which will happen in the near future.

What would be a good name for this repository tesseract_docs, tesseract_documentation, etc?

[gavanderhoorn]

What would be a good name for this repository tesseract_docs, tesseract_documentation, etc?

both seem OK to me.

tesseract_docs is a bit shorter, so might be easier to work with.

[haudren]

I like tesseract_docs as well, it seems to follow a somewhat established practice with the (now defunct) moveit_docs, but also github's own documentation, Laravel's...

[Levi-Armstrong]

The documentation has been moved but not sure if it resolves the issue.

[gavanderhoorn]

It's still there in the history. Removing it entirely will require purging that branch from history and (probably) a force-push to this repository.

[Levi-Armstrong]

I see, so I need to also move the doxygen and benchmark data to the docs repo and then purge the branch. Is that correct?

[marip8]

We just changed CI to push the documentation artifacts to tesseract_docs and I deleted the gh_pages branch, which should help some with the size. We will still need to purge the gh_pages directory from the previous history though to get rid of it permanently