Re: Git mirrors

[Óscar Fuentes]

John Wiegley <address@hidden> writes:

>>>>>> Eli Zaretskii <address@hidden> writes:
>
>>   bzr revision-info address@hidden
>
>> produces output that is easier to parse:
>
>>   96032 address@hidden
>
> Wow.  On my VPS, doing this a single time takes 16 seconds.  Plus, I have to
> do it within the right branch, which means that programmatically, I'll have to
> do everyone within every branch.  That would take a little over a year.
>
> Is there not a better to do this?

As mentioned on my previous post, do a `bzr log --show-ids' and harvest
the revision numbers and revision ids from there. Here, a single `bzr
revision-info ...' takes 8 seconds, but a full `bzr log --show-ids'
takes only 30. Add a few seconds more for retrieving the numbers/ids,
and you are done in less than a minute (which still seems a lot to me,
but YMMV)

> Maybe with some Python code that accesses
> the repository info directly and can turn a file full of rev-info's into the
> corresponding output?  Then I could add a method that only outputs the new
> ones.

This is an interesting path. Hacking the bzr sources is easy and
fun. But this is not the right place for your question. When I
participated there, the guys on the bazaar ml used to be quite
collaborative with this sort of questions.

However, as confirmed by Eli, what you already provided is good enough.