Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
@AtuM I want to either slap you or do homoerotic things to you for saying that. it is yet undecided.
..... why... would... i... want... to... use... a... container...
i... don't... even.. have.. a container system installed... on.. my.. machine. and its a laptop.
does noone see how very very annoying that sounds ?
'well you only had trouble with one lib, so install kerberos on your machine, fuck with that for 2 days to run a lib you may or may not want to use' lol -
You get prebuilt Tensorflow for most linuxes. Did something like "pip install tensorflow" or "conda install tensorflow" not work?
-
@theKarlisK the issue is the source needs minor edits to compile so it imports the correct headers in the correct order
Eg header file references stuff from headers that are included in the cc files so it bombs
Some of
The files during one build stage seem to get deleted if the build fails so I can’t edit them -
@RememberMe I’m afraid it won’t because tensorflow says it only supports Ubuntu which is a Debian variant and I’m using fedora and it’s a big download lol
-
@Root it’s decently usable though since Debian is a damn turtle and I hate Ubuntu and mint doesn’t have gnome 3 by default
-
@Root besides the package source i described has a weird expectation of a compiler
-
so i get the tensorflow build started, and it downloads several hundred mb's of additional binaries and sources which I don't think tensorflow itself ever mentioned, likely so fucking bazel itself will work.
several hundred megs.
this is scary, i have no idea why it downloaded another curl, i have several versions on my system as it is lol and then all these 100+ mb binaries from github bazel build and google wtf ? -
so has someone else been doing all this shit stealing my time period under better circumstances they shouldn'y have pretending to be alive ?
I know you whore shits do that. -
i'm not against efficiency or improvements
but there is a certain repose people used to have at many kinds of jobs that just isnt there
we can afford to slow down on human efficiency, thats what technology is for.
and we should really be prioritizing better, because what I am seeing around me is only efficiently converting my country to a literal waking hell and or nightmare where noone cares about anything LITERALLY. -
@iii its become like instead of just trying to stay out of trouble, they all try to cause it as much as possible.
-
@MadMadMadMrMim several hundred MB is nothing for builds, this isn't a toy to-do list app. Also check if you're doing debug or release builds.
-
@RememberMe I realize this lib is pretty big
But as I said this was mentioned in the docs
It also ended up being cloud extensions I’ll never use and for some reason wouldn’t build anyway
So I disabled them -
@RememberMe why in the fuck is it using boring ssl ???? and why is it trying to mix my source with its source, which is an openssl build !?!
-
@rEaL-jAsE lol
It is using boring ssl though
Tensorflow is pissing me off
Bazel downloads a bunch of shit not in the deps list
And then fails building them -
@AtuM it’s not the os I’m blaming I’m blaming tensorflow
Fedora uses the same build tool chain as every other Unix -
@AtuM dude I have had two Libs out of everyone I’ve been building have unmitigabtable errors from vanilla source
Only two !
I don’t want to hear containers
That concept makes ME vomit
I’m not running a 20 proc 50gb ram machine goddamn it !
And I don’t have the money to rent from some underutilized heat pump of a cloud center either god we’re back in the 1980s -
@AtuM I hate you
I’m looking at the ducking docker guide or I’m going to use pytorch instead
I just need to be able to load goddamn pretrained object classifieds
A package that doesn’t build on a standard tool chain that downloads bunch of extra shit should immediately be a no no
My fucking instincts say NO but my stubbornness says yes! -
@AtuM well here is the question i need an answer to.
so lets say for some goddamn reason tensorflow is dowloading a bunch of fucking shit that it says it doesn't need in the compiling from source section like a fucked up bloated piece of shit project that it is.
so for some reason, despite all reason, when it searches for boringssl it gets openssl's include files first when the gnu preprocessor runs.
how do i make sure that it gets the one it should ?
https://github.com/tensorflow/...
LOOK ! -
@AtuM well code wouldn't build under windows at all like i said lol
you know i'm sitting here trying to get the same prepocessing code working to start training what was my first neural network before they stole all my shit to repress my memory last time and I felt a moment of poignancy as I remembered a little bit of a helping hand indicating why the ImageDraw.getdraw().line function doesn't work
which is getdraw() returns an fing array who's first member defines that code
like that makes any sense
why couldn't he have just been my friend instead of a fucked up monster that endears himself to people and then feeds off them because he's incapable of loving anything ? like most of you people ? -
@AtuM this is kind of an example of one of many things that if given priority people could work on to add variance and purpose back to the world
I believe that somehow you could monetize open source software while still making it free since even major software companies profit from os -
@AtuM i'm not using fedora in production you fool. lol and what should I use debian ?
or ubuntu ?
I guarantee you that if I used either of those I'd have had problems with some of the source somewhere. -
@AtuM your help was suggesting i run containers. my response is derived from similar frustration.
and its everything working minus this one piece.
I even got opencv working and cmake fucks up link txt files across all its modules related to a SINGLE SPECIFIC library, tiff.
I think my question was specific to the compiler suite. how do I make source being included by the build process, which it downloads, be considered over the installed lib I am using.
OpenSSL is pretty standard man.
BoringSSL doesn't seem to be in my experience, but what I do know, I only bounce back and forth between Debian based linux distros and Redhat ones every couple of years.
Yes you're having a bad day, and that's fine.
I'm surrounded by idiotic monsters who think its clever to say they never did something before, when they clearly did, because they're rapists in 'hiding'. -
@rEaL-jAsE this problem really isn’t fixed yet ? Soon there will
Be nothing but back burner shit to fix and no one with the expertise in specific code bases to fix them
We need to monetize fixes somehow -
@rEaL-jAsE honestly the scale of payment would be an issue to be discussed
I think it should be based on numerous factors
Like this one I don’t quite know how to fix unless they add code to specifically use one lib over another which I could swear to god was around at one point but didn’t work worth shit
Maybe someone who knows the code assigns so many hours until people start claiming issues
But then how to reward concurrent efforts or ensure their isn’t too much redundant issue while ensuring the fix gets worked on and just doesn’t sit around
That would be difficult to coordinate
Then you need people to merge and test and write unit tests who know what they’re doing which should all be included in what I believe should be someone scanning for example the GitHub posts who has authority and rewriting them into bug entries
Then what if the fix is crap ? Or how to incentivize making code more optimized ?
Am I just an idiot for using a RedHat variant ?
Why is Bazel such a piece of crap and why do I have to use it to build Tensorflow ??
question