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/
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
EverYoun_ has quit [Ping timeout: 255 seconds]
rtohid has left #ste||ar [#ste||ar]
fane_faiz1 has joined #ste||ar
diehlpk has joined #ste||ar
diehlpk has quit [Remote host closed the connection]
diehlpk has joined #ste||ar
diehlpk has quit [Remote host closed the connection]
diehlpk has joined #ste||ar
K-ballo has quit [Quit: K-ballo]
diehlpk has quit [Ping timeout: 248 seconds]
diehlpk has joined #ste||ar
diehlpk has quit [Remote host closed the connection]
<biddisco>
not bad - we'll get all the PR builds for a given day that way, then parse the xml/htmp and probably get the highest BuildId from it ...
<biddisco>
something like that
<biddisco>
actually, just using the link you generated might be enough for github, it just mean 2 clicks from the user instead of 1
<biddisco>
we use the PR name/date/machine and it will return all the builds that day, then the user just clicks the top one
<heller>
*nod*
<heller>
I think name + machine is even enough
<biddisco>
the filtercount doesn't seem to work if I use date of yesterday I see all the builds, but I think it's fine. I will try it now
<biddisco>
on the github status I mean
<heller>
biddisco: there is also a typo here: " pycicle Config error s0"
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
<biddisco>
ok, I updated the format of the status https://github.com/STEllAR-GROUP/hpx/pull/3042 the results are not real cos I'm just testing the setting and link generation, but now I can do it properly.
<biddisco>
must work on other code now before getting to zurich
<simbergm>
I'm just feeding work to you (because you didn't have anything else to do...) ;)
<github>
[hpx] andreasbuhr opened pull request #3051: Fix HPX-Qt interaction in Qt example. (master...fix_qt_example) https://git.io/vbGHE
mcopik has joined #ste||ar
<simbergm>
heller: I guess you're aware that minimal_timed_async_executor_test doesn't build because of some deferred_call stuff?
<hkaiser>
simbergm: merging #3039 will change that
<simbergm>
ok, nice :)
<hkaiser>
that one fixes a lot of things
<hkaiser>
simbergm: can we add more to #3050?
<simbergm>
hkaiser: yes!
<simbergm>
this is just what I know so far
<simbergm>
you can edit it right?
<hkaiser>
#3039 should go in - or maybe not add just tickets and PRs
<simbergm>
yeah, I'm not sure yet how fine grained the issue should be, I tried to split up the bigger topics but we do whatever seems useful
<simbergm>
but added 3039 now
<hkaiser>
simbergm: we should fix tickets and PRs to be associated with the correct milestone, that gives us a means of tracking as well
<simbergm>
yes, I agree, I'd like to use that as much as possible
<simbergm>
I guess my intent with 3050 was more of a list of definitely broken things
<hkaiser>
k
<hkaiser>
fair enough
<K-ballo>
VS15.5 produces a metric ton of deprecation warnings
<hkaiser>
from HPX?
<K-ballo>
hard to tell.. the wide majority comes from boost
<hkaiser>
you can disable those for HPX at least
<K-ballo>
I'd like to disable them for boost, not for HPX
<hkaiser>
good luck
<hkaiser>
heller: pls see hpx-users
<heller>
hkaiser: saw it
<hkaiser>
tks
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
<hkaiser>
simbergm: who decides now what PRs canbe merged and when?
<hkaiser>
I'd suggest not to let PRs pile up, from experience - this turns into a major pain after a while
eschnett has quit [Quit: eschnett]
<simbergm>
hkaiser: I don't want to be a total dictator about merging things into master, after all you all are much better judges of how a PR will affect other things
<simbergm>
all I wish for is a focus on fixing tests, which I think there is now :)
<simbergm>
is there some particular PR you're thinking about?
fane_faiz1 has quit [Ping timeout: 255 seconds]
<simbergm>
what I want to be a dictator about is merging PRs which break new tests ;) but for that we need the CI sorted out for PRs
<K-ballo>
so how is the CI stuff coming?
<simbergm>
well, CircleCI 2.0 is still an option but has limited resources, biddisco's pycicle thing seems actually very promising now, and tomorrow we will see if CSCS can provide something usable with jenkins
<simbergm>
one of those three should be merged as soon as possible, but I don't yet know which
<hkaiser>
simbergm: if needed we can invest some money into circleci
eschnett has joined #ste||ar
hkaiser has quit [Quit: bye]
biddisco has quit [Ping timeout: 240 seconds]
hkaiser has joined #ste||ar
<simbergm>
hkaiser: that (investing in circleci) would be nice, I believe that CSCS could also invest something, although there is nothing official
fane_faiz1 has joined #ste||ar
<simbergm>
I think we should see how our meeting goes tomorrow about jenkins, and then discuss the alternatives and see what the pros/cons are of each service
<hkaiser>
k
<heller>
I think either circleci or hyper.sh look good
<heller>
hyper.sh with buildbot_travis that is
<simbergm>
I guess the question with circleci is if it will be able to replace *all* of the current builds/testing (and then some)
<simbergm>
because it doesn't support remote workers, right?
<hkaiser>
simbergm: do we have to exclusively use just one CI service?
biddisco has joined #ste||ar
<simbergm>
hkaiser: no, I guess not
<simbergm>
but I think if we can get for example rostam/cscs to build and test PRs there shouldn't be much need for circleci (the way it's used now)
<hkaiser>
it could still do the header checks
<hkaiser>
those require quite some time
<biddisco>
hkaiser: now that pycicle can set the build status of PRs I am thinking that we can just deploy one at LSU, one at FAU and one ant CSCS and it can do everything we need.
<biddisco>
we can also do inspect and other checks
<hkaiser>
k
<simbergm>
hkaiser: yes, that's possible, I just imagine that LSU + CSCS + FAU should have enough cpu hours to do all the checks we might want to do
<hkaiser>
we should keep appveyor, though
<simbergm>
agree with appveyor
<biddisco>
appveyor is windows right?
<simbergm>
yep
aserio has joined #ste||ar
<hkaiser>
and we should possibly revive a mac builder
<biddisco>
cscs might have some old macs we can use ...
<hkaiser>
k
<biddisco>
I am thinking about dedicated raspberry pi for pycicle now .... :)
<hkaiser>
we might have one as well
<biddisco>
good
<hkaiser>
biddisco: great! will help with your chipmunk problem ;)
<biddisco>
lol
<hkaiser>
simbergm: I think we should go ahead with merging one of #3033, #3035, #3047, #3051, or #3021 once it passes all tests
<hkaiser>
simbergm: in my experience, one PR a day keeps things at a managable pace
<simbergm>
hkaiser: I have nothing against that :)
<simbergm>
although now we should be able to check some of them on cdash as well
<simbergm>
biddisco: we can't trigger rebuilds with pycicle yet right? some commits have all tests failing...
<hkaiser>
simbergm: pls go ahead - you're the master of ceremonies now ;)
<biddisco>
if you merge one, greina will rebuild the remaining PRs, then if the red gets less, merge the one with fewest fails, repeat and rinse
<simbergm>
hkaiser: oh, ok, thanks :)
<simbergm>
biddisco: good idea
<biddisco>
daint is playing up at the moment and not displaying build/test results.
<biddisco>
after I added the HPX_RUNNER=srun thingy that heller reminded me about
<hkaiser>
simbergm: let me know if you need admin access to the repo to merge things
akheir has quit [Remote host closed the connection]
<simbergm>
hkaiser: looks like I don't have the rights to merge
<hkaiser>
sec
<hkaiser>
you should have now
<simbergm>
hkaiser: yep, thanks!
<hkaiser>
simbergm: no pushing to master anyways ;)
<simbergm>
heh, no way ;)
<simbergm>
I think I could be called a hypocrite in that case...
* zao
-f
EverYoung has joined #ste||ar
aserio has quit [Quit: aserio]
aserio has joined #ste||ar
EverYoung has quit [Ping timeout: 252 seconds]
aserio has quit [Read error: Connection reset by peer]
aserio has joined #ste||ar
eschnett has quit [Quit: eschnett]
eschnett has joined #ste||ar
eschnett has quit [Client Quit]
biddisco has quit [Ping timeout: 240 seconds]
<heller>
hkaiser: we just need more PRs that fix tests, and have the PRs flagged as "proper" before merging, then things won't pile up ;)
<K-ballo>
"proper" as in "approved"?
<heller>
proper as in: we know it doesn't break any new tests?
<heller>
+ being approved
biddisco has joined #ste||ar
<heller>
biddisco: hmm, you might need to allocate more than one node?
<heller>
Some tests need 4 localities
<heller>
-N4 -n1 could be enough as well
<heller>
In the batch script
rod_t has joined #ste||ar
rod_t has left #ste||ar [#ste||ar]
rod_t has joined #ste||ar
eschnett has joined #ste||ar
aserio has quit [Remote host closed the connection]
aserio has joined #ste||ar
rod_t has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rod_t has joined #ste||ar
simbergm has quit [Ping timeout: 260 seconds]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
rod_t has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
david_pfander has quit [Ping timeout: 255 seconds]
rod_t has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
aserio has quit [Ping timeout: 255 seconds]
rod_t has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rod_t has joined #ste||ar
fane_faiz1 has quit [Ping timeout: 276 seconds]
rod_t has quit [Client Quit]
rod_t has joined #ste||ar
rod_t has quit [Client Quit]
rod_t has joined #ste||ar
rod_t has quit [Client Quit]
rod_t has joined #ste||ar
biddisco has quit [Ping timeout: 240 seconds]
aserio has joined #ste||ar
rod_t has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
parsa[[[w]]] has quit [Read error: Connection reset by peer]