Opened 15 years ago
Closed 15 years ago
#355 closed task (fixed)
skip test (rather than avoiding them)
Reported by: | Peter Johansson | Owned by: | Peter Johansson |
---|---|---|---|
Priority: | trivial | Milestone: | svndigest 0.7 |
Component: | build | Version: | trunk |
Keywords: | Cc: |
Description (last modified by )
related to ticket:362
For example tests that need the test repository should not be avoided by AM_CONITIONAL, but instead they should return 77 (which signals SKIP) to make check
Change History (8)
comment:1 Changed 15 years ago by
Description: | modified (diff) |
---|
comment:2 follow-up: 3 Changed 15 years ago by
My understanding of svn export is that the directory structure is exported without svn meta information. If we work in a svn export'ed environemnt we should skip tests? Also, what is the difference with 'HAVE_SVN_WC' and that repository is available. I thought these say the same thing.
comment:3 Changed 15 years ago by
Replying to jari:
My understanding of svn export is that the directory structure is exported without svn meta information.
Exactly.
If we work in a svn export'ed environemnt we should skip tests?
No what I suggest is that tests should not be skipped in an svn export'ed environment.
Also, what is the difference with 'HAVE_SVN_WC' and that repository is available. I thought these say the same thing.
At time being HAVE_SVN_WC
is set to false if svnversion returns 'exported', in other words, if svn meta information is not available. So HAVE_SVN_WC
means we have svn meta information, which we don't have when building from a dist or from an 'svn export'. OTOH, if you export from the repository you will get everything checked in, including the test repository. And if the repository is there, we should use it in tests.
Of mote, HAVE_SVN_WC
is not ill defined because the inital purpose of it was to decide whether we should rebuild lib/svnversion_info.cc
(which issues svn info
). The mistake was to use HAVE_SVN_WC
in test/Makefile.am
as having svn meta info and having access to the test repository is not the same thing.
comment:4 Changed 15 years ago by
comment:5 Changed 15 years ago by
comment:6 Changed 15 years ago by
Milestone: | svndigest 0.x+ → svndigest 0.7 |
---|---|
Status: | new → assigned |
comment:7 Changed 15 years ago by
comment:8 Changed 15 years ago by
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
Given ticket:362, it is clear that the check if we
HAVE_SVN_WC
is not the most relevant way to decide if we should skip some tests. Readsvn export
. We should check if the repository is available instead.