[Python-Dev] Looking for VCS usage scenarios

Brett Cannon brett at python.org
Tue Nov 4 03:40:52 CET 2008


On Mon, Nov 3, 2008 at 14:38, Gustavo Niemeyer <gustavo at niemeyer.net> wrote:
>> Both. First and foremost I am looking for any scenarios people are
>> using now for svn that I didn't cover. After that I can probably add
>> some DVCS-specific things. But the problem with that is my DVCS
>> experience is limited and thus I don't want to add a scenario that
>
> So try to listen to people that actually use these tools and are
> trying to help you.
>

I am and why I will add either Thomas' or Barry's scenario.

>> seems whizbang cool but in real life is never used; premature
>> optimization and all that.
>
> As I mentioned early in my mail, all the scenarios I described were
> specifically meant to expose common situations that we go through in
> our workflow.  I'm not picking weird things to try favoring any
> specific tool, even because I believe all tools should support these
> very basic desires.
>

I didn't think you were picking scenarios that were odd, but I am sure
you also have an IT staff to make sure your workflow can support all
of that (see my reply to Barry).

>> As I said, one of the scenarios already says patches can be whatever
>> the DVCS supports the best; plain diffs, branches, etc. And the
>> comments for that scenario will point out any perks from that feature.
>
> """
> This scenario to meant to represent the steps required for a non-core
> developer who has a one-off patch they want to create from a read-only
> checkout.
> """
>
> Ok, so would you mind to expand that to describe that the tool should
> send the history of the "one-off patch", and what it means to not send
> it?
>

"Send" as in automatically submit?

> Otherwise you're really comparing oranges and apples.  If I send you
> "svn diff" or "bzr diff", you get a plain diff.  Merging this will not
> carry the history of who made the chances, what were the incremental
> steps, and will handle conflicts without context, which may be way
> more boring to perform than when you actually have full history
> information.  That's not the same as using "bzr send".
>

Hey, if what ``bzr send`` generates can be uploaded to
bugs.python.org, then that is what will be listed in the scenario.

> Again, that's a real world situation, not something I'm making up to
> favor bzr.  I'm pretty sure hg should have something similar too (and
> git does for sure).
>
>> As for the "via mail" feature, I use Gmail so it doesn't mean anything to me. =)
>
> Please ask Tim to teach you how the "mail" word got in "Gmail".  ;-)

I have learned to ignore my uncle for years so it won't do much good. =)

-Brett


More information about the Python-Dev mailing list