[Replicant] Status of the scripts to migrate to Mediawiki Was: DEPRECATED previous patches given inexperience related mistakes

[Denis 'GNUtoo' Carikli]

On Tue, 25 Jan 2022 13:59:40 +0000
Alexis via Replicant <replicant at osuosl.org> wrote:

> I would like to advise you to deprecate all my previous patches
> regarding wiki-migration-scripts given they had major issues withing
> commit messages or within patches themselves, sorry for such
> mistakes, but is only by doing so that one learns, this is relating
> my inexperience with git, hopefully I will avoid this mistakes in the
> future.
The issue is that the code status on my side wasn't great, so that
complicated things a lot: 
- my work was in an unmerged branch, and some of the patches were
  hackish attempts to fix bugs. While it made sense to do that to try
  to understand the bugs, it complicated a lot the rebases.
- The code in master and in the unmerged branch also contained many
  bugs.

I've now fixed part of the issue by spending time to cleanup my
unmerged branch, to rebase it on top of master and to finally merge it.
I've also reduced the amount of bugs on master.

While master most probably still contains bugs, we have at least a
robust base to sent patches against and collaborate more easily.

Right now, at least importing both the libsamsung-ipc and the Replicant
infrastructure wiki in git work.

As for the main Replicant wiki we still have a Redmine bug to get fixed
in Redmine and/or to workaround for it to work, so I've not tried that
yet with the new code.

We also need to verify that everything went well, and at that point we
can start testing the migration to real Mediawiki instances.

Note that the current plan is to migrate the textile data as-is:
- First we'd use the migration scripts to add all the wikis to git,
  without modifying the content. The goal of this step is only to keep
  the real history of the wiki pages because it contains very important
  information (like references to where the information added to the
  wiki comes from).
- Then we could convert the pages either manually or automatically.
  While saving the retrieved textile to git.replicant.us and rewriting
  the history to convert it to mediawiki syntax is possible, I think
  it's easier to not modify the git history and instead add git commits
  that convert pages to the Mediawiki syntax. There are many advantages
  for that:
  - It keeps the real history. This lowers the risk of loosing data due
    to conversion bugs.
  - It enable people to manually do this conversion if they need to as
    once imported in a Mediawiki instance we'd see the textile syntax
    when clicking on edit.
  - We could also convert it with tools and push the changes with git
    or even manually in Mediawiki by editing the page and using copy
    and paste.
  - We can see the textile syntax in the history of the page,
    especially when comparing versions, so it should be user friendly
    enough.
  Because of that I don't think that changing the format of Redmine
  wikis before the migration is a good idea since it would increase the
  risk of having issues, and we might also loose the real history along
  the way if the code that does the migration converts all the pages
  revisions/versions to the new format.

Note that we need to see how Mediawiki reacts with files that ends
in .textile instead of .mw. If not we'll have to rewrite a bit the
history and rename the files but at least that should not touch the
files content, so the risk of having issues is low here.

Denis.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.osuosl.org/pipermail/replicant/attachments/20220125/f3fe9463/attachment-0001.asc>