rcurtin_irc changed the topic of #mlpack to: mlpack: a scalable machine learning library (https://www.mlpack.org/) -- channel logs: https://libera.irclog.whitequark.org/mlpack -- NOTE: messages sent here might not be seen by bridged users on matrix, gitter, or slack
<TarekNasser[m]> <zoq[m]> "> <@marcusedel:matrix.org..." <- I noticed that the dataset was already uploaded but without the labels so i used it and added the labels manually.
_slack_mlpack_19 has joined #mlpack
_slack_mlpack_22 has joined #mlpack
<jonpsy[m]> Hey, how do I figure and debug error in `Azure pipeline`. I'm reviewing this [PR](https://github.com/mlpack/mlpack/pull/3164). Previously I used to Appveyor into windows build, not sure how to deal with Azure. Any help?
<zoq[m]> jonpsy[m]: Unfortunately you can't ssh into the azure machine, so either you can reproduce the error locally or you have to rely on the output.
<jonpsy[m]> Ah that sucks!
<jonpsy[m]> MSVC is notorious for its errors, its really hard to debug. How do you guys manage?
<jonpsy[m]> a lot of times, windows has errors you can't reproduce anywhere else except MSVC. That too that specific MSVC ver. 🤷‍♂️
<rcurtin[m]> yeah, you are absolutely correct. I tend to search the error logs for "error " (the space is important) to figure out what is going on. the most typical error is a C1060 (out of RAM) which mostly I don't worry about and hope that things like #2777 and removing Boost will fix 😃
<jonpsy[m]> i see. Yeah it does cost a lot of RAM
<jonpsy[m]> how do I re-run it?
<rcurtin[m]> I think the azure pipelines interface has a button on the left sidebar that allows re-running
<rcurtin[m]> (I'm not a fan of the huge number of clicks required to do basically anything in the azure pipelines interface...)
<rcurtin[m]> yeah, so there is a "re-run" button on the build page: https://github.com/mlpack/mlpack/pull/3146/checks?check_run_id=5730467515
<EshaanAgarwal[m]> rcurtin[m]: I think this option is only available to members or maintainers. I am also facing MSVC errors in my PR and I can't see a re-run option there.
<rcurtin[m]> yeah, it should be only for folks who are a part of the organization; but, whoever is reviewing your PR can restart it if needed (or if it is an 'expected failure' then we can just not worry about it. you've probably noticed that very few merged PRs have a fully green status... it's hard to keep everything building correctly!)
<jonpsy[m]> Okay I'll do it
<jonpsy[m]> Lo n behold! As suspected..
<jonpsy[m]> I can't seem to find any errors besides this. Does this mean its GTM?
RajatMishra[m] has quit [Quit: You have been kicked for being idle]
<zoq[m]> <jonpsy[m]> "I can't seem to find any..." <- I guess at this point we are just ignoring the windows build and hope once boost is refactored out we don't end up with a bunch of windows specific issues.
<jonpsy[m]> Cool, I'll approve that PR then
<jonpsy[m]> but if we should also bookmark the PR's we're merging which stands on this assumption. That of "it'll work when boost fixes"
<zoq[m]> I think rcurtin wanted to take a look at it as well, it's about the matrix size check isn't it?
<jonpsy[m]> aye
<jonpsy[m]> I think from my side its gtm, so i'll approve from my side. We'll merge when rcurtin approves too (and not with the automatic approval). Sounds good?
<zoq[m]> jonpsy[m]: Sounds good to me, I mean we have the system in place that people can just leave a comment and say I'd like to look at it please wait.
<zoq[m]> zoq[m]: So I guess if there is no comment, from anyone no need to wait.
<rcurtin[m]> yeah, no particular need to wait on me, but I'll try to get to it today
<GopiMTatiraju[m]> zoq: rcurtin have a look whenever you get time.
<GopiMTatiraju[m]> * have a [look, * look](https://gitlab.com/conradsnicta/bandicoot-code/-/issues/10#note_893421085) whenever
byperion has joined #mlpack
byperion has quit [Client Quit]