On 22/06/2009, at 09.53, Florian Weimer wrote:
> Why haven't those patches been applied yet? I think I can guess the
> reason: To me, it's still unclear what type of limit is best. A
> per-diff limit doesn't actually bound the message size, so it's
> problematic. If set too low, it also affects single-file changes, and
> if set too high, it doesn't bring down the size of large merges.
> Truncating the whole message instead is somewhat rude, and makes sense
> only as a safety net to protect your mail infrastructure.
Why not apply all of them? And then let the users decide which they
want to use?
> What about keeping track of the line count so far, and after a limit
> is exceeded, replace the diffs (and only the diffs) with a line like
> this one:
>
> [21 added lines, 18 deleted lines in 3 hunks omitted]
>
> As a result, you get 5 lines and about 350 bytes per changed file.
> It's technically still unbounded, but even for fairly large merges (or
> loads), the total message size will remain somewhat manageable.
>
> I think this might be a good compromise among the trimming options.
> Thoughts?
I think it is a good idea.
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2364109
Received on 2009-06-22 11:01:35 CEST