[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: 'svn diff' vs 'svnlook diff'

From: Julian Foad <julianfoad_at_btopenworld.com>
Date: Wed, 21 Nov 2012 00:21:37 +0000 (GMT)

Johan Corveleyn wrote on 2012-11-19:

> Julian Foad wrote:
>> Johan Corveleyn wrote:
>>> Daniel Shahaf wrote:
>>>> Johan Corveleyn wrote:
>>>>> I currently have a patch sitting here for adding
>>>>> --diff-cmd to 'svnlook diff',
>>>>
>>>> I wonder what's the minimal change we could make to the cmdline
>>>> client such that it can operate on transactions (and thus void
>>>> the need to reimplement every svn proplist/diff/cat/info switch
>>>> in svnlook).  (Read-only, at least initially.)
>>>>
>>>> Is this something Julian's tree-read-api branch would address?
>>
>> Yes my tree-read-api work would make this sort of thing easier.
>>
>>>> Maybe we need to implement svn_ra_local_txn (like ra_local, but
>>>> with HEAD being a transaction instead of a revision)?  Other ideas?
>>
>> Move the core tree-diffing functionality down a layer from
>> libsvn_client into libsvn_diff.  Let 'svn' pass some kinds of 'tree
>> description' inputs to it (from the WC and RA interfaces) and let
>> 'svnlook' pass other kinds of 'tree description' inputs
>> (revisions and txns, from the repos layer).
>
> Am I right in thinking that this is something that will probably take
> some time to complete (i.e. not for 1.8)?

Yes.

> I'm just wondering whether I should commit my patch for adding
> --diff-cmd to 'svnlook diff'. If you (or anyone else) intend to
> "merge" both diff drivers soonish, it doesn't make much sense to
> work specifically on --diff-cmd.
>
>>> I'd like to note that the output of 'svnlook diff' is slightly
>>> different from 'svn diff', and I'd like to preserve that different
>>> behavior (or at least preserve the svnlook behavior here). IMO the
>>> output of 'svnlook diff' is better suited for post-commit emails.
>>
>> Ugh.  Most of these differences are (IMO) unwanted.  I basically
>> agree with your comments below about which ones are better.
>
> Agreed. Would be nice indeed if both implementations could converge,
> and take on the best format.

I am doing a bit of de-duplication in this area now.  I'm not sure how far I'll take it.

- Julian
Received on 2012-11-21 01:22:14 CET

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.