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

Re: MD5 backward compat code?

From: Erik Huelsmann <ehuels_at_gmail.com>
Date: 2005-12-09 12:20:29 CET

On 12/9/05, Malcolm Rowe <malcolm-svn-dev@farside.org.uk> wrote:
> On Fri, Dec 09, 2005 at 10:23:35AM +0100, Erik Huelsmann wrote:
> > I'd like to remove the fallback from svn_wc_transmit_text_deltas(),
> > also because it is a backward compat to pre-1.0 working copies. I
> > think the time to update those to post 1.0-working copies has elapsed
> > :-)
> >
> Sounds fine to me - we are talking about something that was changed in
> r4374, pre-0.17. maxb already removed some of this support in r16042.

Great! :-)

> I have been wondering to what extent we really need to treat absent or
> all-zero checksums as valid as well, since that also seems to be for
> (nebulous) 'compatibility'. That's probably a separate change, though.

There are some cases in which there are still no MD5s in the entries
file even though you'd expect them to be. To test this, I put an
abort() in write_entries(). The test suite started failing big time.
I'd like to address this, but I'm working on perf tuning first,
because it fits the 'libsvn_wc performance release 1.4' so well.
Depending on how long it takes for 1.4 to be released, I may get
another wc-format bump going to require md5s.


Received on Fri Dec 9 12:25:39 2005

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.