#62 closed defect (fixed)
builds which are intended to use the embedded copy of Crypto++ might get their headers from a system-supplied copy of Crypto++ — at Version 3
Reported by: | midnightmagic | Owned by: | |
---|---|---|---|
Priority: | major | Milestone: | |
Version: | 0.5.19 | Keywords: | |
Cc: | Launchpad Bug: |
Description (last modified by zooko)
As you can see here:
buildbot-pycryptopp/builders/MM%20netbsd5%20i386%20warp/builds/39/steps/test/logs/stdio
... it looks like with a system crypto++ installed (in my case, 5.6.1) there are pieces pulled in from the crypto++ 5.6.1 header files which reference a CryptoPP:: function which does not exist in the embedded version, which therefore causes the test to fail.
The fix appears to be to tell (as I suspect you were going to anyway) setup.py to disable the embedded cryptopp stuff.
Change History (3)
comment:1 Changed at 2010-11-16T07:01:34Z by zooko
- Summary changed from Add --disable-embedded-cryptopp to warp buildslave to builds which are intended to use the embedded copy of Crypto++ might get their headers from a system-supplied copy of Crypto++
comment:2 Changed at 2010-11-16T07:15:05Z by zooko
- Resolution set to fixed
- Status changed from new to closed
fixed by 79a464530fa3357da9a57f29f39d0ad63211880c/git. See buildbot-pycryptopp/builders/MM%20netbsd5%20i386%20warp/builds/39 (before, bundled Crypto++) versus buildbot-pycryptopp/builders/MM%20netbsd5%20i386%20warp/builds/40 (after, bundled Crypto++) and buildbot-pycryptopp/builders/MM%20netbsd5%20i386%20warp%20syslib/builds/29 (before, system-provided Crypto++) versus buildbot-pycryptopp/builders/MM%20netbsd5%20i386%20warp%20syslib/builds/31 (after, system-provided Crypto++).
Thanks for the buildslave and the bug report!
comment:3 Changed at 2014-07-22T18:49:04Z by zooko
- Description modified (diff)
The problem is that the build which was intended to use the embedded copy of Crypto++ is getting the header file from the system instead of from the bundled copy.