aserio changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/
diehlpk has joined #ste||ar
diehlpk has quit [Ping timeout: 248 seconds]
parsa has quit [Quit: Zzzzzzzzzzzz]
Smasher has quit [Remote host closed the connection]
<github>
hpx/master 24d7e3e Thomas Heller: Merge pull request #3065 from STEllAR-GROUP/workaround-3058...
<heller>
jbjnr: it would be nice if we could retrigger all open PRs that are mergeable when a new commit to master was made
<heller>
To always reflect the latest state.
parsa has joined #ste||ar
<heller>
jbjnr: you also need to build the plugins, the parcel coalescing for example
<heller>
jbjnr: regarding my TMI complaint. It's actually both, too much and too little. If you look at the current dashboard, we have three builds looking exactly the same but having different failures and start time. Are all three important?
<jbjnr>
heller: all prs are rebuilt when master is changed already
<jbjnr>
yes of course they are all important. if master was changed 3 times, then there are going to be 3 different builds for master, and if the errors change between them we want to see it
<jbjnr>
in general, just make sure that you click the build time list and sort by build time, newest at top, it will remember this setting between views, then all is good
<jbjnr>
and the 3 builds are 549,553,555 and +4,+2 etc, they even show you which tests are passing in the next build that didn't pass in the one before
<jbjnr>
if you don't like it - just use buildbot. I'm not forcing you to look at cdash
<jbjnr>
I find the results much more usable on cdash than on buildbot. look at buildbot this morning, jut a sheet of orange, with no clues as to what is wrong.
<jbjnr>
on cdash I can see exactly that k-ballo commit one fx to the build stuff, and 4 build errors went away. marveloous - this is exactly what I want to see.
<heller>
jbjnr: give me some time to get used to it ;)
<heller>
Rebuild when master changes: very good
<heller>
I can see that now
<heller>
Getting there
<heller>
I think the two missing failures are now due to some missing libraries
<jbjnr>
if a test needs parcel coaescing, then surely the cmake dependencies shoud trigger it to be built - if it is not being built -then it means the test is not setting its dependencei correctly I'd say ...
<heller>
Yeah
<heller>
It's currently being built as part of the "core" target
<heller>
If that's not invoked explicitly your out of luck
<jbjnr>
tests should depend on core?
K-ballo has joined #ste||ar
<heller>
jbjnr: I guess so, yes
parsa has quit [Quit: Zzzzzzzzzzzz]
parsa has joined #ste||ar
parsa has quit [Client Quit]
<heller>
jbjnr: I'll try to come up with a different UI tomorrow. With cdash giving me the info