Opened 6 years ago
Closed 6 years ago
#5264 closed defect (fixed)
Coverage history plots are broken
Reported by: | Francesco Casella | Owned by: | Adrian Pop |
---|---|---|---|
Priority: | high | Milestone: | 1.14.0 |
Component: | Testing Framework | Version: | |
Keywords: | Cc: | Martin Sjölund |
Description
All the .svg plots underhttps://libraries.openmodelica.org/branches/history/
are broken since the start of the year, see, e.g., https://libraries.openmodelica.org/branches/history/newInst/Modelica_trunk.svg.
Please restore them to their original state.
Change History (4)
follow-up: 2 comment:1 by , 6 years ago
comment:2 by , 6 years ago
Replying to adrpo:
I'm afraid that is my doing.
There are now two coverage tests for newInst, one ran on libraries.openmodelica.org (to test against master) and one run on ripper.openmodelica.org.
Aha, I was not aware of the former one.
The one on ripper has the full history, the one on libraries only has the history from December. Depending on when you look at the history plots (after ripper is finished or after libraries is finished) you will either get the full history or not.
That also explains why there are now partially overlapping history reports here.
Honestly, though it's not such a big deal, I find both things very confusing. BTW, it's not only newInst
tests, see, e.g., https://libraries.openmodelica.org/branches/history/master/Modelica_StateGraph2.svg
I guess we could remove the coverage from libraries and leave only the one on ripper, but then we would not get the comparison with master (the newInst will be removed from the tables here):
https://libraries.openmodelica.org/branches/overview-combined.html
What about putting the results of libraries in another directory, e.g. https://libraries.openmodelica.org/something-appropriate/?
comment:3 by , 6 years ago
Cc: | added |
---|
I now removed newInst branch from the coverage ran on libraries.openmodelica.org.
The problem is that we run some of the things on one computer and other things on another computer and this way we cannot merge the results as the databases build are local.
We need to change to a network based database which would make sharing the results rather easy. This way you can distribute even more on several computers (of course the timings would be un-comparable).
comment:4 by , 6 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
I'm afraid that is my doing.
There are now two coverage tests for newInst, one ran on libraries.openmodelica.org (to test against master) and one run on ripper.openmodelica.org. The one on ripper has the full history, the one on libraries only has the history from December. Depending on when you look at the history plots (after ripper is finished or after libraries is finished) you will either get the full history or not.
I guess we could remove the coverage from libraries and leave only the one on ripper, but then we would not get the comparison with master (the newInst will be removed from the tables here):
https://libraries.openmodelica.org/branches/overview-combined.html