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

Re: "Checksum mismatch" when performing update

From: Matt Doran <matt.doran_at_papercut.biz>
Date: 2005-05-11 14:27:10 CEST

I'm not using a proxy ... and network communication is over the loopback
interface ... so that's probably not the prob. But I've done some
digging and found something pretty weird.....

    $ svn info providers.sln
    Path: providers.sln
    Name: providers.sln
    Repository UUID: 505ffd8f-26e6-0310-8e5c-c33a79a468e7
    Revision: 251


    Checksum: 8f21805e9b084c2ce6cbbb52f325272e

So the files revision is 251....

    $ diff providers.sln .svn/text-base/providers.sln.svn-base
    $ md5sum providers.sln .svn/text-base/providers.sln.svn-base
    8f21805e9b084c2ce6cbbb52f325272e providers.sln
    8f21805e9b084c2ce6cbbb52f325272e .svn/text-base/providers.sln.svn-base

The WC and text-base file are identical ....

    $ svn cat -r251 providers.sln | md5sum
    7dfee8873e30d2f574a79b7e77fb9418 -
    $ svn cat -r251 providers.sln > /tmp/test.251
    $ diff providers.sln /tmp/test.251
    < !!!!

But the wc and pristine copy are different than the repository ... a
single line containing 4 '!' chars.

I can believe that I might have accidentally added the "!!!!" line the
the wc file ... but not the text-base or the checksum in 'entries' to
match!? That's weird! Almost too strange to believe.

Anything stupid I could have done to cause this?

(sorry if I've wasted your time)


Ben Collins-Sussman wrote:

> On May 11, 2005, at 3:37 AM, Matt Doran wrote:
>> $ svn --version -q
>> 1.1.4
>> $ svn up
>> svn: Checksum mismatch for 'providers.sln'; expected:
>> '9cb298b2150e77340280d646c50d244a', actual:
>> '6590988692d8eb0e8ed5514b5821cd79'
>> $ md5sum providers.sln
>> 8f21805e9b084c2ce6cbbb52f325272e providers.sln
>> $ md5sum .svn/text-base/providers.sln.svn-base
>> 8f21805e9b084c2ce6cbbb52f325272e .svn/text-base/providers.sln.svn- base
> The checksum error isn't necessarily about the text-base. Checksums
> are sent in both directions over the network, whenever binary diffs
> are sent in 'update' or 'commit'.
> So it's possible that an incoming binary diff got munged over the
> network...
Received on Wed May 11 14:44:07 2005

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

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