diff options
| author | Robert Managan <managan1@llnl.gov> | 2008-10-08 16:30:07 (GMT) |
|---|---|---|
| committer | Robert Managan <managan1@llnl.gov> | 2008-10-08 16:30:07 (GMT) |
| commit | 08bf8fbd0b18f1737ac101f3eba31d41b8c9fb5c (patch) | |
| tree | f2db9aab60c73c1c5476854985f7ef49cbe2394f /test/Java | |
| parent | f50c85585ecd15a547257d311ff953d59fd94d09 (diff) | |
| download | SCons-08bf8fbd0b18f1737ac101f3eba31d41b8c9fb5c.zip SCons-08bf8fbd0b18f1737ac101f3eba31d41b8c9fb5c.tar.gz SCons-08bf8fbd0b18f1737ac101f3eba31d41b8c9fb5c.tar.bz2 | |
Testing on a Mac turned up a problem that did not show up on linux.
SK and I agree that we are surprised by that.
Both bld = env['BUILDERS']['PDF'] and bld = env['BUILDERS']['DVI'] had
lines like
bld.add_action('.tex', LaTeXAuxAction) # for DVI
bld.add_action('.tex', PDFLaTeXAuxAction) # for PDF
run from the initialization of two tools. This was a BAD thing.
That cured some test failures. For the rest I made all the TEX tests
initialize the environment with the os.environ PATH since the check
for the existence of the tools with where_is that looks on the system path.
So the tests still quietly exit if the tool is not on the system path
and now find if it is in a non-standard location like I have.
Diffstat (limited to 'test/Java')
0 files changed, 0 insertions, 0 deletions
