Re: [Tails-dev] [review'n'merge:1.1] feature/tor-launcher-0.…

Supprimer ce message

Répondre à ce message
Auteur: anonym
Date:  
À: The Tails public development discussion list
Sujet: Re: [Tails-dev] [review'n'merge:1.1] feature/tor-launcher-0.2.5.4
26/05/14 11:40, intrigeri wrote:
> Hi,
>
> anonym wrote (18 May 2014 22:33:47 GMT) :
>> This branch imports Tor Launcher 0.2.5.4 (from 0.2.5.3) + a few fixes
>> from master (see commit message). With this package, our delta against
>> upstream has only decreased since upstream has fixed several things we
>> patched in our Tor Launcher. So beside merging in the new upstream
>> changes, I've just reverted our old fixes in favour of upstream's.
>
> Woo, congrats!
>
> May you please update #6840 ("Upstream our Tor Launcher changes") so
> that one can get its current status without having to read the entire
> comments thread? I suggest creating a subtask for each bugfix or
> feature that was not merged upstream yet.


Done.

>> For the review, the commit in Tails' Git isn't very interesting, since
>> it corresponds to the imported package.
>
> Slightly OT (and we already had a thread about it, see #7036):
>
> I'm a bit sad not to see progress on #7087 ("Do not bundle Tor
> Launcher in the main Tails Git repository"). Three months ago, back
> when this was merged for 0.23, our agreement was that this was a very
> temporary thing, only justified by the bridges support work being
> completed super-late in that release cycle, and 1.0 being
> a point-release.
>
> OTOH, I'm still unsure what the best way would be to resolve this
> problem (the more general ticket about it being #7036), so I really
> can't blame you for this :) I was once sold to submodules, but later
> read that they have a whole bunch of problems of their own, e.g.
> in the merges handling. I'm more and more convinced this general
> problem could be an important one to tackle for 2.0 or 3.0, as part of
> making our workflow a bit smoother.


The thing is that I trust you will find a much better solution than me.
Do you think the Git submodule approach still is worth switching to now,
even if that also may end up as a temporary solution?

>> The relevant changes in our Tor
>> Launcher git are 7fa3b70..2ebf1db. The non-reverts introduced by "Tails
>> developers" are patches that were upstreamed; the revert:s (and merge of
>> ttp/masteR) by "Tails developers" are the only things I've done in our
>> repo. Sorry that this has become this confusing...
>
> Indeed, it's getting very hard (especially when combined with how
> #6840 looks) to get what the status of our delta is. I'd rather not
> see this get worse and worse any longer. May you please save the
> history of our current master branch (e.g. as "old-master"), and
> rebase it on top of current upstream's master, dropping everything
> that was merged upstream, the corresponding reverts, and all
> merge commits?
>
> Once this is done, and the status of #6840 is updated, I'm happy to
> merge this branch. It looks good to me, and my quick testing didn't
> expose any issue.


Done. `git diff old-master..master` showed me that there was some
" " inconsistencies left from a94a6d7 (or possibly some incorrect
merge conflict resolution after it). I updated the version shipped in
Tails' Git so that unintentional delta is also eliminated.

Cheers!