[PATCH 17 of 21 WIP-PY3K] obsolete: use xrange from py3kcompat

[Matthew Turk]

On Mon, Oct 12, 2015 at 3:10 PM, Matt Mackall <mpm at selenic.com> wrote:
> On Mon, 2015-10-12 at 16:25 -0500, Matthew Turk wrote:
>> # HG changeset patch
>> # User Matthew Turk <matthewturk at gmail.com>
>> # Date 1444684321 18000
>> #      Mon Oct 12 16:12:01 2015 -0500
>> # Node ID 97c6284c8ba29a8c595e3bdcbb858e91e34332d1
>> # Parent  1842237016d2b252a9060941a850f54b90451b9e
>> obsolete: use xrange from py3kcompat
>>
>> diff -r 1842237016d2 -r 97c6284c8ba2 mercurial/obsolete.py
>> --- a/mercurial/obsolete.py   Mon Oct 12 16:11:19 2015 -0500
>> +++ b/mercurial/obsolete.py   Mon Oct 12 16:12:01 2015 -0500
>> @@ -78,6 +78,7 @@
>>      phases
>>  )
>>  from .i18n import _
>> +from .py3kcompat import xrange
>
> A few comments on this series.
>
> First, please don't send 21 patches. It's bad for overall list flow
> control. Aim for something more like 5.
>
> Second, this individually importing of symbols is going quickly get to
> be a headache. We should just have:
>
>  from .compat import *  # things get ugly when you break the whole world
>
> ..and put anything we need for cross-version compatibility there. We've
> already got some of this in util.py.
>
> Third, please don't do 50 trivial conversions like xrange or cStringIO
> one file per changeset. Just do them all in one go: "global: replace all
> cStringIO users".
>
> Patch 16 has two unrelated changes. Don't do this, it's not a shortcut.
> Here the second change is highly questionable, so it'd be likely to sink
> the first.. if it wasn't already policy to reject such changes.
>
> Lastly, the conversion to absolute imports should go in its own series
> so it's not blocked on other concerns.

Thanks, Matt!  I'll keep working, update the series when ready, and send a V2.

>
> --
> Mathematics is the supreme nostalgia of our time.
>