[Bug 3650] New: hg archive -S with svn subrepo: images become corrupted

[bugzilla-daemon at bz.selenic.com]

http://bz.selenic.com/show_bug.cgi?id=3650

          Priority: normal
            Bug ID: 3650
                CC: mercurial-devel at selenic.com
          Assignee: bugzilla at selenic.com
           Summary: hg archive -S with svn subrepo: images become
                    corrupted
          Severity: bug
    Classification: Unclassified
                OS: Linux
          Reporter: thomas at intevation.de
          Hardware: PC
            Status: UNCONFIRMED
           Version: unspecified
         Component: Mercurial
           Product: Mercurial

With Mercurial 2.3.2, svn 1.6.12 on Debian/squeeze:

Mercurial repo with svn subrepo included via https:
base = [svn]https://svn.wald.intevation.org/svn/mpuls/base/trunk/

base/mpulsweb/public/images/logos/ contains some images, they are marked as
binary, e.g.:
$ svn propget svn:mime-type
base/mpulsweb/public/images/logos/logo_intevation.png
application/octet-stream

Size is 2383 bytes.

$ hg archive -t files -S ../archive -I
base/mpulsweb/public/images/logos/logo_intevation.png

The result is 2382 bytes (one byte less)

original starts with:
 89 50 4E 47 0D 0A 1A 0A  00 00 00 0D 49 48 44 52  .PNG.... ....IHDR
archived starts with:
 89 50 4E 47 0A 1A 0A 00  00 00 0A 49 48 44 52 00  .PNG.... ...IHDR.

i.e. 0D-0A is changed to 0A, just 0D is changed to 0A.

subrepo.py uses "svn cat" to access the file, I noticed the following:
ssh hostname "svn cat
/PATH/TO/wasku/base/mpulsweb/public/images/logos/logo_intevation.png" > x1
ssh -t hostname "svn cat
/PATH/TO/wasku/base/mpulsweb/public/images/logos/logo_intevation.png" > x2

x1 is identical to the original file.
x2 is 2391 bytes long and starts with:
 89 50 4E 47 0D 0D 0A 1A  0D 0A 00 00 00 0D 49 48  .PNG.... ......IH
i.e. 0A is changed to 0D-0A

With
"svn export
/PATH/TO/wasku/base/mpulsweb/public/images/logos/logo_intevation.png"
it does not matter if ssh is used with or without -t, the result is good in
both cases.


I see two ways to solve this:
1. find a way to make "svn cat" behave as expected
2. use "svn export" to access the file data (probably needs a temporary file
   if not already archiving to files)

Suggestions?

-- 
You are receiving this mail because:
You are on the CC list for the bug.