Just for completion my reply once again for the list (replied to Bert only by 
accident)
 
Thx for your quick response. Unfortunately your linked version 1.8.3.1 doesn't 
fix the problem for me (tried both 32 and 64bit). Though your description of 
the circumstances leading to the fixed zlib problem really sounded like a "yay 
that must be it" ;o).
 
For an addition - all 1.7.x clients work as expected with the same server 
setup/repository.
 
Are there any specific printouts from the server side which I could gather from 
our IT for helping to hunt it down?
 
Well - whenever you need to test something you can come back to me. I am able 
to reproduce that problem with a hit count of 100% ;). For now I downgraded to 
1.7.11 for being able to work again.
 
Best Regards,
Christian Steineck

Analytik Jena AG | R&D Software Analytical Instrumentation| 
Konrad-Zuse-Straße 1 | 07745 Jena | Germany |
Phone: +49 (0) 36 41  77-87 53 | Fax: +49 (0) 36 41  77- 92 79|
E-Mail: c.stein...@analytik-jena.de | Internet: www.analytik-jena.de |
 
Von: Bert Huijben [mailto:b...@qqmail.nl] 
Gesendet: Dienstag, 17. September 2013 12:57
An: 'Christian Steineck'; dev@subversion.apache.org
Betreff: RE: Regression in 1.8.x - checksum mismatch when using svn via 
http/apache
 
                Hi,
 
If everybody could reproduce this as easy as you describe here we probably had 
this fixed years ago. As I can't check your specific server setup I will try to 
help you by just guessing.
 
I'm guessing that this is the 'zlib' problem we finally identified last week.
 
Can you verify if the refreshed binaries (that should be available on 
http://sliksvn.com/pub/Slik-Subversion-1.8.3-1-win32.msi and 
http://sliksvn.com/pub/Slik-Subversion-1.8.3-1-x64.msi) soon resolve your 
problem?
 
If this is the case it is caused by a specific combination of zlib compilation 
options, a specific file contents and bad luck on the size of your network 
packets.
 
                Bert
 
PS.
Until the files are available there you can temporarily find them at:
http://b.qqn.nl/f/2013/tmp/Slik-Subversion-1.8.3-1-win32.msi
http://b.qqn.nl/f/2013/tmp/Slik-Subversion-1.8.3-1-x64.msi
 
 
From: Christian Steineck [mailto:c.stein...@analytik-jena.de] 
Sent: dinsdag 17 september 2013 10:41
To: 'dev@subversion.apache.org'
Subject: Regression in 1.8.x - checksum mismatch when using svn via http/apache
 
Hi there,
 
when updating to current svn version 1.8.x I spotted a regression.
 
Used environment:
On client side:
-          Windows 7 SP1 Ultimate 64bit
-          SlikSVN 1.8.0 (also tested with SlikSVN 1.8.3 and TortoiseSVN 1.8.3 
- all behave the same) - 64Bit Version
-          Also tried SlikSVN 1.8.0 32Bit Version - same problem
 
On server side:
-          Apache/2.4.6 (Unix) PHP/5.4.19 SVN/1.8.3
 
Reproduction recipe:
-          Checkout a big project (in my test case 64MB of data) via http (svn 
checkout http://svnserver/repo/project/trunk)
 
Expected behavior:
-          The checkout completes and the working copy has the complete project 
data
 
Detected behavior:
-          Checkout fails on random file with svn: E200014: Checksum mismatch
-          The mismatch is always on project files not on the .svn/ metafiles
 
If you need more information I will try to request them from our IT.
 
 
Best Regards,
Christian Steineck

Analytik Jena AG | R&D Software Analytical Instrumentation| 
Konrad-Zuse-Straße 1 | 07745 Jena | Germany |
Phone: +49 (0) 36 41  77-87 53 | Fax: +49 (0) 36 41  77- 92 79|
E-Mail: c.stein...@analytik-jena.de | Internet: www.analytik-jena.de |
 

Analytik Jena AG mit Sitz in Jena | HR-B 200027 Amtsgericht Jena | Umsatzsteuer 
ID Nr.: DE 150509863
Aufsichtsratsvorsitzender: Andreas Krey | Vorstand: Klaus Berka (Vorsitz) | 
Jens Adomat, Stefan Doehmen

Analytik Jena AG mit Sitz in Jena  |  HR-B 200027  Amtsgericht Jena  |  
Umsatzsteuer ID Nr.: DE 150509863
Aufsichtsratsvorsitzender: Andreas Krey  |  Vorstand: Klaus Berka (Vorsitz) | 
Jens Adomat, Stefan Doehmen

Reply via email to