Opened 11 years ago
Closed 7 years ago
#9521 closed defect (duplicate)
Cliquer: Move test commands from spkg-install to spkg-check
Reported by: | mpatel | Owned by: | tbd |
---|---|---|---|
Priority: | minor | Milestone: | sage-duplicate/invalid/wontfix |
Component: | packages: standard | Keywords: | |
Cc: | ncohen, leif | Merged in: | |
Authors: | Reviewers: | Jeroen Demeyer | |
Report Upstream: | N/A | Work issues: | |
Branch: | Commit: | ||
Dependencies: | Stopgaps: |
Description
From Cliquer's spkg-install
:
if [ "$SAGE_CHECK" = "yes" ]; then echo "Compiling and running the test cases of cliquer..." make testcases if [ $? -ne 0 ]; then echo "Failed to compile test cases of cliquer... exiting" exit 1 fi ./testcases if [ $? -ne 0 ]; then echo "Failed to run test cases of cliquer... exiting" exit 1 fi fi
We can move this code (or most of it) to a new spkg-check
.
Since SAGE_LOCAL/bin/sage-spkg
contains
cd $BASEDIR if [ "$SAGE_CHECK" != "" -a -f spkg-check ]; then echo "Running the test suite." chmod +x spkg-check ./spkg-check if [ $? -ne 0 ]; then
we don't need to check SAGE_CHECK
in spkg-install
.
Change History (7)
comment:1 Changed 10 years ago by
comment:2 Changed 10 years ago by
- Cc ncohen leif added
I've closed #9767 as a duplicate of this one. I've cc'ed Leif, who commented on #9767 and Nathann as the package maintainer.
I need to respond to Leif's comment on #9767 though:
Dave, be warned: You'll hate me if I should review this (spkg-install
, patches/Makefile
).
I think Leif has unreasonable expectations some times - in particular #9603, (an iconv change), has resulted in a large number of changes, which all took time, which could have been better spent addressing other more serious issues in Sage. The number of changes Leif requested on #9603 was excessive I feel. Six weeks after the ticket was created to just add
&& [ "x$UNAME" != xHP-UX ]
it is still awaiting a positive review.
I'm not denying the changes Leif requested on #9603 has improved the iconv package - though I think it was one of the better packages to start with. I do however think the time spent on it could have been put to more productive use.
If tickets to add a very small change are going to result in weeks of iterations, then the changes are not going to be made - at least not by me.
I am happy to attempt to clean up Cliquer, but on the understanding it does not become like #9603. I've no desire to spend so much time on endless small changes.
The only changes I would make to patches/Makefile
are Solaris-specific, as the compiler flags are wrong. (Oh, and I might add the necessary change to package so it builds on HP-UX, where shared libraries are .sl and not .so.) But I have no intension of re-writing the whole of the Makefile.
There's a lot of unnecessary things in spkg-install
and I'm happy to remove them, but I'm not willing to spend weeks on it.
Dave
comment:3 Changed 10 years ago by
comment:4 Changed 8 years ago by
- Milestone changed from sage-5.11 to sage-5.12
comment:5 Changed 7 years ago by
- Milestone changed from sage-6.1 to sage-duplicate/invalid/wontfix
- Reviewers set to Jeroen Demeyer
- Status changed from new to needs_review
Duplicate of #9870.
comment:6 Changed 7 years ago by
- Status changed from needs_review to positive_review
comment:7 Changed 7 years ago by
- Resolution set to duplicate
- Status changed from positive_review to closed
See
http://groups.google.com/group/sage-devel/browse_thread/thread/88102b4c4eadbba7/0100c924f44cf18c?hl=en
I think in this case, it is better to run the test code in
spkg-install
on every build, as it takes takes very little time - about 5 seconds to do so on my machine. As such, I suggest weOne issue here is that
make
exits with 0, even if the test suite fails. I found this after purposely inducing some failures. It makes me wonder how many other test suites do this too!See also #9767 and some comments from myself and Leif.
Dave