UmbralRaptor changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> … one of the other grad students just compared me to nomal O_o | <ferram4> I shall beat my problems to death with an engineer.
Snoozee is now known as Majiir
e_14159 has quit [Ping timeout: 204 seconds]
e_14159 has joined #kspacademia
<UmbralRaptor>
Oh, right. I should source more Touhou music. >_>
<SnoopJeDi>
we have an IT audit of sorts tomorrow, I wonder how that might affect us if it suddenly struck >_>
<SnoopJeDi>
42 inane questions about every single system
Technicalfool has joined #kspacademia
Majiir is now known as Snoozee
<UmbralRaptor>
So, oh, do people use conda because macs are bad at package management and take heroic efforts to keep your python libraries out of date?
<UmbralRaptor>
s/oh/uh/
<Qboid>
UmbralRaptor meant to say: So, uh, do people use conda because macs are bad at package management and take heroic efforts to keep your python libraries out of date?
icefire has quit [Read error: Connection reset by peer]
<SnoopJeDi>
Conda's definitely got it's flaws but I'd rather things proceed on the path Continuum has been blazing than the tarpit that pip/virtualenv/whatever the hell else have been stuck in for...ever.
<whitequark>
continuum isn't "blazing" anything, they produce software that barely works and changes under your feet regularly
egg|phone|egg has joined #kspacademia
<SnoopJeDi>
I guess your experience with conda has been more pathological than mine, then
<egg|phone|egg>
!Seen UmbralRaptor
<Qboid>
egg|phone|egg: I last saw UmbralRaptor on [21.04.2017 15:04:15] in #bottorture saying: "аррӏеapple"
<egg|phone|egg>
!Wpn UmbralRaptor
* Qboid
gives UmbralRaptor a ██████████████ bird
<whitequark>
SnoopJeDi: it has been a never-ending stream of problems and nothing else
<SnoopJeDi>
I don't agree
<whitequark>
I'm not actuallye exaggerating when I'm saying "it's worse than exchanging tarballs"
<whitequark>
it IS worse than exchanging tarballs.
<whitequark>
and it's not like we don't know how to make package managers, either.
<SnoopJeDi>
The package management part of conda needs a lot of work
<SnoopJeDi>
But I think conda-forge is promising enough
<whitequark>
SnoopJeDi: it has been [19h] *flips* [0h] since a completely spurious new conda issue appeared
<whitequark>
this is the fourth one just this week
<whitequark>
yeah, tell me about it.
<SnoopJeDi>
¯\_(ツ)_/¯
<SnoopJeDi>
If it "barely works" in your opinion I'm interested in how you'd explain the userbase it has. Is it all hype?
<whitequark>
the userbase are academics.
<whitequark>
it's par for the course.
<whitequark>
you just need low enough standards to consider it acceptable.
<SnoopJeDi>
I don't agree with that either
<SnoopJeDi>
If anything, academics seem to underadopting it compared to the data science types (who have their own pathologies)
<SnoopJeDi>
at least the ones I'm rubbing shoulders with, which is about as biased a sample as you can get
<whitequark>
it's not so much "hype" as "network effects"
<whitequark>
*we* are using conda solely because of network effects
<SnoopJeDi>
I use it because it works
<SnoopJeDi>
personally
<SnoopJeDi>
and for my use case
<whitequark>
consider yourself extremely lucky
<SnoopJeDi>
I don't, though
<SnoopJeDi>
I think it's far more likely your use case is just stuck in corner land
<whitequark>
it's not
<whitequark>
I don't have an use case
<whitequark>
I'm literally just packaging a few Python things and a few C libraries
<whitequark>
this is smack in the middle of the thing conda is supposed to do
<whitequark>
and the failures (for the most part) have nothing to do with our code or projects in particular
<SnoopJeDi>
Package creation does sound like a goddamned nightmare
<SnoopJeDi>
Recipes and blah blah
<whitequark>
that's a part of conda. conda-build.
<SnoopJeDi>
Yes, I know
<whitequark>
but for example the most recent failure is a dependency resolution issue.
<whitequark>
quite remarkably the exact same code worked yesterday
<SnoopJeDi>
I can't imagine packaging for it is very pleasant at all, no
<SnoopJeDi>
I'll keep using it until someone makes something better, which is inevitable
<SnoopJeDi>
You're right that the bar is pathetically low in academia, though. Fermilab looked at the wide selection of ways to build software, and decided "nah let's roll our own" https://cdcvs.fnal.gov/redmine/projects/contractor
<whitequark>
-_____-
<SnoopJeDi>
Bonus points awarded for the fact that if you follow the installation instructions for the development version of Synergia and clone their repo and run the install script, the first thing the install script does is...try to run a `git` command
<SnoopJeDi>
(which was not available on the machine I wanted to install on at the time, and it was a real hassle to make it available)
<SnoopJeDi>
But hey, at least there *is* documentation and a process rather than "idk lol I needed feature X so I smashed it into the codebase, got it to work just enough for my one-off use, did not document it, and now it's the basis for 3 other features written by other people"
* SnoopJeDi
glares at OPAL
<SnoopJeDi>
I'm writing what will hopefully turn into upstream documentation as I learn how the hell it works, but it's more an act of petty rage than of actual desire to be helpful
<SnoopJeDi>
I wish nothing but mild inconveniences on the developers responsible for my headaches. May their barista fill their coffee cup to the very brim!