|
Poopernickel posted:turns out python is good actually
|
#
?
Oct 21, 2020 21:42
|
|
|
|
| # ? Apr 23, 2024 15:36 |
|
|
I sometimes dip into other languages but honestly, Python is really nice to come home to. Dealing with venvs is garbage, but most of what I write is covered by the stdlib and what OS doesn't have requests or flask in its distro packages? So what if it's not the newest, I just make sure that my stuff works on Debian stable, deal with it. Another upside: no makefile / automake / meson / ninja / cargo / …
|
|
#
?
Oct 21, 2020 21:54
|
|
|
starbucks hermit posted:C++ is too sexy. I've used it plenty of times earlier in my career and it's not bad, but I have a couple projects that could use multithreading and I don't want to engineer another project around pthreads for now. c++11 onwards has a fairly slick wrapper around pthreads, transforms mutex locks into the RAII pattern, which fits reasonably well and guarantees you don't forget to drop locks (they get dropped automatically when they go out of scope) i found it quite good to work with, compared to raw pthreads API
|
|
#
?
Oct 22, 2020 03:56
|
|
|
Antigravitas posted:I sometimes dip into other languages but honestly, Python is really nice to come home to. Dealing with venvs is garbage, but most of what I write is covered by the stdlib and what OS doesn't have requests or flask in its distro packages? So what if it's not the newest, I just make sure that my stuff works on Debian stable, deal with it. venv when containers exist????
|
|
#
?
Oct 22, 2020 03:57
|
|
|
C is only good language
|
|
#
?
Oct 22, 2020 04:01
|
|
|
Celexi posted:C is only good language c not only gives you the loaded gun, but puts your finger on the trigger and points the gun at your foot.
|
|
#
?
Oct 22, 2020 04:02
|
|
|
BobHoward posted:c++11 onwards has a fairly slick wrapper around pthreads, transforms mutex locks into the RAII pattern, which fits reasonably well and guarantees you don't forget to drop locks (they get dropped automatically when they go out of scope) Sounds similar to the idea of smart pointers, which can attach heap allocation to scope. You still have to allocate and manage threads, though, and do semaphores intelligently, and etc. Nevertheless, it does sound better than wrestling with raw pthreads.
|
|
#
?
Oct 22, 2020 05:33
|
|
|
ratbert90 posted:venv when containers exist???? the entire python community had multiple decades to figure out packaging and deployment and nothing they came up with is ever preferable to a docker image based on Debian stable. it’s pathetic
|
|
#
?
Oct 22, 2020 05:43
|
|
|
Nomnom Cookie posted:the entire python community had multiple decades to figure out packaging and deployment and nothing they came up with is ever preferable to a docker image based on Debian stable. it’s pathetic Almost every application that has multiple dependencies and even a slight amount of complexity is better distributed via a container. This is not a Python specific issue. edit There’s a Python docker image based on alpine which is much smaller than Debian stable.
|
|
#
?
Oct 22, 2020 05:52
|
|
|
ratbert90 posted:Almost every application that has multiple dependencies and even a slight amount of complexity is better distributed via a container. This is not a Python specific issue. I agree with you that containers are great and most things should use them. what’s impressive about python is that the native packaging is always worse than a docker image. it will occasionally happen that distributing a go binary or a fat jar is the right choice, but you never, ever want to be shipping wheels or whatever the gently caress python uses this year ps. alpine sucks
|
|
#
?
Oct 22, 2020 06:00
|
|
|
Nomnom Cookie posted:ps. alpine sucks alpine has the one benefit of earning you a bonus when you double performance by replacing it with something that isn't trash
|
|
#
?
Oct 22, 2020 06:27
|
|
|
Alpine might be trash but it’s better than Deb “hold my beer while I fix this SSL warning without knowing what’s going on” Ian.
|
|
#
?
Oct 22, 2020 07:07
|
|
|
what’s wrong with alpine? I never looked into more than reading description that says its just a small base image
|
|
#
?
Oct 22, 2020 07:20
|
|
|
It's optimized for size, which makes it slow. If you replace it with for example debian you can expect performance gains, which in large deployments can be significant.
|
|
#
?
Oct 22, 2020 07:25
|
|
|
aside from dropping all optimizations in favor of size at all costs, musl also has some hilariously bad decisions around networking that aren't repeated anywhere else pretend i went on a 3 page rant about musl's DNS implementation querying all resolv.conf nameserver entries in parallel every time and using the result from whichever entry responded first at random
|
|
#
?
Oct 22, 2020 07:31
|
|
|
ratbert90 posted:venv when containers exist???? That's a venv but also for C programs. venvs are bad.
|
|
#
?
Oct 22, 2020 07:33
|
|
|
ratbert90 posted:Alpine might be trash but it’s better than Deb “hold my beer while I fix this SSL warning without knowing what’s going on” Ian. alpine simply doesn't have a security update channel at all
|
|
#
?
Oct 22, 2020 07:36
|
|
|
Also that SSL issue was over a loving decade ago.
|
|
#
?
Oct 22, 2020 07:40
|
|
|
spankmeister posted:Also that SSL issue was over a loving decade ago. the problem with being the best example of why rng should be taken seriously also makes it hard for people to forget such an incredible blunder.
|
|
#
?
Oct 22, 2020 08:00
|
|
|
spankmeister posted:Also that SSL issue was over a loving decade ago. they’ve made gently caress all changes to their process to stop a recurrence and still make Ill advised changes to packages because they know better than the devs
|
|
#
?
Oct 22, 2020 08:57
|
|
|
Progressive JPEG posted:aside from dropping all optimizations in favor of size at all costs, musl also has some hilariously bad decisions around networking that aren't repeated anywhere else Is there a write up of this dns behaviour anywhere ?
|
|
#
?
Oct 22, 2020 08:59
|
|
|
idk about a writeup but it's glibc and it's the reason for "options single-request"
|
|
#
?
Oct 22, 2020 09:34
|
|
|
jre posted:Is there a write up of this dns behaviour anywhere ? musl maintains an extensive list of incompatibilities but most of the others in the list do not result in indeterminate behavior that varies depending on how your network is feeling that day
|
|
#
?
Oct 22, 2020 09:40
|
|
|
Linux Developers Discussing Possible Kernel Driver For Intel CPU Undervolting
|
|
#
?
Oct 22, 2020 13:01
|
|
|
there’s also msr_safe to whitelist msrs but kernel devs refuse to upstream it and want drivers written for each processor/platform feature you want to manipulate instead
|
|
#
?
Oct 22, 2020 14:50
|
|
|
If you want to use a fatter container use Centos8 instead of Debian.
|
|
#
?
Oct 22, 2020 16:12
|
|
|
use the ubi images that you get support for under your organization's rhel licenses, op
|
|
#
?
Oct 22, 2020 16:49
|
|
|
i'd rather use perl than python
|
|
#
?
Oct 22, 2020 18:04
|
|
|
hifi posted:i'd rather use perl than python Encouraging self-harm??? Mods?!
|
|
#
?
Oct 22, 2020 18:08
|
|
|
hifi posted:i'd rather use perl than python i am here to start the perl survivors group python (3, really 3.6+) is much better than perl, and it's here to help you move on
|
|
#
?
Oct 23, 2020 00:44
|
|
|
unless the reason you like perl is job security, then more power to you
|
|
#
?
Oct 23, 2020 00:45
|
|
|
one core issue with python is that you had to specify versions twice, going to a recent minor version, to make the recommendation. and that's for a language that's been around since the late 80s.
|
|
#
?
Oct 23, 2020 00:48
|
|
|
Cybernetic Vermin posted:one core issue with python is that you had to specify versions twice, going to a recent minor version, to make the recommendation. and that's for a language that's been around since the late 80s. you couldve said one could only recommend it for the past 4 years of its 40 year history, which is fair as a critique, versioning is a project founder piñata
|
|
#
?
Oct 23, 2020 00:51
|
|
|
Cybernetic Vermin posted:one core issue with python is that you had to specify versions twice, going to a recent minor version, to make the recommendation. and that's for a language that's been around since the late 80s. 3.6 is like 4 years old now, it’s the baseline version anywhere you should be starting new projects at all.
|
|
#
?
Oct 23, 2020 00:55
|
|
|
oh yeah, starting new projects, that thing we all do all day every day. old software being things that just fade off in the distance.
|
|
#
?
Oct 23, 2020 00:58
|
|
|
if you’re not starting a new project then it’s irrelevant which languages are good and which are bad, because you’re already stuck with whatever language the project uses.
|
|
#
?
Oct 23, 2020 01:11
|
|
|
I wrote new Perl way past its prime once purely because Perl's DBI library is mature and the Python postgres client library situation was a shitshow
|
|
#
?
Oct 23, 2020 01:14
|
|
|
Soricidus posted:if you’re not starting a new project then it’s irrelevant which languages are good and which are bad, because you’re already stuck with whatever language the project uses. ok, to just spell this out instead: python has had a horrible track record of backwards compatibility. it is not a bad bet to avoid it for that reason, if you wish for any permanence in the things you put your time into. people have raised packaging always having been bad, but i think that is very much downstream from the core being uninterested in compatibility and practical deployment, and every abstraction away from that you get it gets worse and worse.
|
|
#
?
Oct 23, 2020 01:17
|
|
|
serious q: i do a lot of infra dev at work and we're pretty happy with python 3.6 for that. you need a lovely little daemon to pipe metrics from elasticsearch to datadog or something, write 200 LOC of python and pop it in your kube cluster. easy peasy. what language is superior to python for this application?
|
|
#
?
Oct 23, 2020 01:28
|
|
|
|
| # ? Apr 23, 2024 15:36 |
|
|
Nomnom Cookie posted:serious q: i do a lot of infra dev at work and we're pretty happy with python 3.6 for that. you need a lovely little daemon to pipe metrics from elasticsearch to datadog or something, write 200 LOC of python and pop it in your kube cluster. easy peasy. what language is superior to python for this application? probably java for elasticsearch
|
|
#
?
Oct 23, 2020 01:35
|
|