|
prisoner of waffles posted:my half-assed guess: they want to invest in korean industry and there's essentially 0 way to do that without directly or indirectly investing in / betting on chaebol? sometimes it is better to just admit your limitations not all things can be invested in
|
# ? May 23, 2018 15:50 |
|
|
# ? Jun 4, 2024 13:13 |
Helicity posted:most python devs i know dont give a poo poo about security or performance, so i dont see the problem with using python 2 forever it’s worse than 3 and its library support is decaying proportionally to mass adoption of python 3 also it’s not the latest so many are just upset
|
|
# ? May 23, 2018 16:08 |
|
Comedy option: use Perl
|
# ? May 23, 2018 16:13 |
|
Helicity posted:most devs i know dont give a poo poo about security or performance fixed anyway, I have a hard time getting excited about the py2 to py3 thing. I mean, the handling of it has not been ideal, but on the other hand its likely companies with lots of money and developers who feel they have to be on py2 will come up with a fork that continues to be patched or some other sort of solution....and I have hard time getting excited about the problems big companies are having. if its something small, then its not a big deal to migrate to python 3 if its big and you or your company does not have the resources for the migration, you can ride on the coattails of whatever the big companies are doing granted, there's a window in the range of possibilities where all the big companies say "gently caress it" and migrate to python 3 or do something thats kept in-house or some other thing that doesn't help the little guy with big un-migratable projects. but i'm not going to get excited about that until it happens (ok i won't get excited about it then either because lol getting excited at this stupid poo poo) (also, i feel like the python org learned a lot from the py3 debacle and any future big changes like that will be handled better)
|
# ? May 23, 2018 17:10 |
|
if you want a py3 horror story that's actually relevant, look no further than mercurial. it took multiple years for the test runner to even start up, because of the amount of baggage involved in the port, and that's after multiple custom source code formatter tools were written to add b'' prefixes, etc. to various things. mercurial is a good example of what should be a shoe-in fit for python: a moderately high level system-level tool that deals with files and networking. hg basically has to do undo the damage of py3's unicode model in a file they call "pycompat.py": https://www.mercurial-scm.org/repo/hg/file/tip/mercurial/pycompat.py which implements a py2 str-alike notable ridiculous things, which i have talked about at length before include: python 3's developers decided that filenames and command line arguments should be unicode, because they are unicode on windows. but they aren't unicode on windows, they're UCS-2, and filenames can contain invalid or unpaired surrogates. last i tried, this basically caused python to emit an exception when calling os.listdir() if it contained a file with an unpaired surrogate. i don't know if they've since fixed that. but mercurial needs the original *bytes* of filenames, since that's part of the repo format and the wire protocol, and python unhelpfully makes that hard. I believe you can get bytes filenames if you call os.listdir with a bytestring path, which is just this weird overloading convention thing that makes no sense as a concept, since bytes are supposed to be meaningless by definition! command line arguments are a different variation of the same topic, but related since filenames are often passed on the command line. i still think if you're handed a file object, you might not be able to do f.read().split('\n') safely, since the file object might secretly be in 'bytes' mode, and bytestrings don't have a split method, and the only way to detect whether you're in bytes mode is the incredibly clunky f.read(0) == b''.
|
# ? May 23, 2018 18:40 |
|
Thermopyle posted:(also, i feel like the python org learned a lot from the py3 debacle and any future big changes like that will be handled better) my favorite so far was "guido doesn't like the concept of 3.10 as a version number so after 3.9 we'll go to python 4 with no breaking changes". i don't know if they thought it through at all because /usr/bin/python4 is kind of a huge change unless they're going to have /usr/bin/python3 which is actually python 4. https://www.curiousefficiency.org/posts/2014/08/python-4000.html#comment-1547343622
|
# ? May 23, 2018 18:45 |
|
Suspicious Dish posted:my favorite so far was "guido doesn't like the concept of 3.10 as a version number so after 3.9 we'll go to python 4 with no breaking changes". i don't know if they thought it through at all because /usr/bin/python4 is kind of a huge change unless they're going to have /usr/bin/python3 which is actually python 4. i already have separate /usr/bin/python3.4 and /usr/bin/python3.5 on my pc this may work better than we imagine only because python was pre-emptively stupid for us?
|
# ? May 23, 2018 18:48 |
|
Suspicious Dish posted:hilarious things yeah i agree thats all hilarious but i think its a different sort of hilariousness. (but you might not have been saying it was the same and just pointing out other hilarious things) Suspicious Dish posted:my favorite so far was "guido doesn't like the concept of 3.10 as a version number so after 3.9 we'll go to python 4 with no breaking changes". i don't know if they thought it through at all because /usr/bin/python4 is kind of a huge change unless they're going to have /usr/bin/python3 which is actually python 4. im not sure this means much
|
# ? May 23, 2018 19:05 |
|
wait that very blog post you linked says that subsequent discussions on python-dev concluded that the release after 3.9 will be 3.10 as is right and proper
|
# ? May 23, 2018 19:10 |
|
til: guido is 62yo dude aged like an asian person, wtf
|
# ? May 23, 2018 19:12 |
|
yeah with all the "sabotage" talk i thought there was something deliberately introduced into 2 that broke things with some flimsy 3 justification, not some weird euphemism for "they didn't make it trivial to port 2 to 3" which seems to be the core complaint?
|
# ? May 23, 2018 19:13 |
|
qhat posted:Comedy option: use Perl alive and well. It does what it's intended to do. That being said, i'm as much instructor as i am script kiddie these days, so I do things in py or ruby so these kids can have some takeaway knowledge. i'm fully aware that perl chops are only worth money when somebody needs a dick to stick in some legacy peanut butter
|
# ? May 23, 2018 19:15 |
|
the most beautiful thing to me was being shown "the master VCL" which is what controls the baseline operation of our Varnish, and.....it's got huge blocks of ASM. Nothing else is fast enough
|
# ? May 23, 2018 19:16 |
|
I did a lot of heavy lifting in Perl in my old job, it's not as bad as people make it out to be. I can't really ever see myself using it over python3 nowadays though.
|
# ? May 23, 2018 19:34 |
|
qhat posted:I did a lot of heavy lifting in Perl in my old job, it's not as bad as people make it out to be.
|
# ? May 23, 2018 19:45 |
|
JawnV6 posted:yeah with all the "sabotage" talk i thought there was something deliberately introduced into 2 that broke things with some flimsy 3 justification, not some weird euphemism for "they didn't make it trivial to port 2 to 3" which seems to be the core complaint? they want python 2 to go away irrespective of any other stakeholders' feelings on the matter. because python 2 offends their loving ~aesthetics~ or whatever. there are interested parties who, surprisingly enough, do in fact want python 2 to continue, so that "2.8" fork is now called "tauthon" apparently. i'm not opposed to people making breaking changes as long as they are reasonably infrequent, but the key condition there is that those who inflict breaking changes upon the world do not get to set the timetable. c#, java, react are all projects that understand this. linux poo poo, with the honorable exception of the actual kernel, do not; see python, gtk for the most egregious examples.
|
# ? May 23, 2018 19:56 |
|
Thermopyle posted:if its big and you or your company does not have the resources for the migration, you can ride on the coattails of whatever the big companies are doing The big problem is that it's not easily migratable because we use a large number of libraries to do some quite low-level things with coroutines and asyncio and such, and some of those haven't been updated to python3 yet, or possibly never will be. There might be alternatives but that's the whole point, fixing it is a bunch of work We do that because it's a reasonably fast web interface for an embedded device which can support a number of different protocols with little work, pretty much everything that's used on web including SOAP and CORBA (IoT ). I did not decide any of that I also write the whole stack underneath including C/C++ applications, kernel drivers and firmware. The ones that aren't outsourced anyway VVVVV also that, we only move kernel when we really have to. Still better than some of the devices which we resell which run 2.6.17 or something (thank god for Private Speech fucked around with this message at 20:23 on May 23, 2018 |
# ? May 23, 2018 20:11 |
|
Sapozhnik posted:they want python 2 to go away irrespective of any other stakeholders' feelings on the matter. because python 2 offends their loving ~aesthetics~ or whatever. The kernel breaks modules all the time, I mean if you consider recompiling poo poo all time as stable then yeah but binary compatibility is not a thing
|
# ? May 23, 2018 20:11 |
|
Jonny 290 posted:alive and well. It does what it's intended to do.
|
# ? May 23, 2018 20:16 |
|
i give zero fucks about ruby vs python vs perl. they're all fine, and they're all garbage, and for the same reasons anyone who has a strong opinion about one of those being better or worse than the other two is someone whose opinions i care less about in the future
|
# ? May 23, 2018 20:25 |
|
Gazpacho posted:it was intended to generate awk reports over multiple files lol perl does this very well
|
# ? May 23, 2018 20:25 |
|
Suspicious Dish posted:my favorite so far was "guido doesn't like the concept of 3.10 as a version number so after 3.9 we'll go to python 4 with no breaking changes". i don't know if they thought it through at all because /usr/bin/python4 is kind of a huge change unless they're going to have /usr/bin/python3 which is actually python 4. I kind of like Python 3, but this is totally ridiculous. Well, I guess it just means I won't like Python 4.
|
# ? May 23, 2018 20:43 |
|
Notorious b.s.d. posted:perl does this very well
|
# ? May 23, 2018 20:50 |
|
Jonny 290 posted:i'm fully aware that perl chops are only worth money when somebody needs a dick to stick in some legacy peanut butter pm me the rest of this extended metaphor
|
# ? May 23, 2018 21:14 |
|
prisoner of waffles posted:pm me the rest of this extended metaphor it gets really bad when they start making sandwiches
|
# ? May 23, 2018 21:17 |
Suspicious Dish posted:i still think if you're handed a file object, you might not be able to do f.read().split('\n') safely, since the file object might secretly be in 'bytes' mode, and bytestrings don't have a split method, and the only way to detect whether you're in bytes mode is the incredibly clunky f.read(0) == b''. https://docs.python.org/3/library/stdtypes.html#bytes.split That said, I think you are still right because you would need to call f.read().split(b'\n') for the bytes and f.read().split('\n') for the str. And the criticism that file paths should be considered opaque bytes is 100% accurate IMO. e: I think Rust gets this right, by having separate types for unicode strings (String), platform-specific strings (OsString), and paths (Path, which is a wrapper around OsString). The three types can all be converted to each other simply enough, but you are forced to explicitly handle cases where that conversion fails. VikingofRock fucked around with this message at 21:57 on May 23, 2018 |
|
# ? May 23, 2018 21:45 |
|
Athas posted:I kind of like Python 3, but this is totally ridiculous. Well, I guess it just means I won't like Python 4. they're not doing what suspicious dish claims there. at one point Guido said that, but then they thought about it and decided its a bad idea. so, no the version after 3.9 is not 4.0, it's 3.10 (at least thats what is claimed at that link) so yes Guido has bad ideas, but no theyre not following this particular bad idea
|
# ? May 23, 2018 22:10 |
|
i find to find the mailing list thread where it was discussed and shot down (because i remember hearing about it much more recently than that), but got somewhat lost in the giant thread about using bytes for file paths. i didn't even mean to do that it just appeared. guess bad ideas are forever. https://mail.python.org/pipermail/python-dev/2014-August/
|
# ? May 23, 2018 22:51 |
|
i thought you were right because I would've sworn I heard about it more recently than 4 years ago too, but as far as i could find its 3.10 after 3.9 i've googled across more recent discussions saying 4.0 was after 3.9 but they're all just quoting old stuff that is apparently not true anymore i guess we'll see!
|
# ? May 24, 2018 00:56 |
|
im surprised no one mentioned the big announcement weve all been waiting for: http://micronaut.io/announcement.html
|
# ? May 24, 2018 01:12 |
|
Malcolm XML posted:yeah me too. stay safe tef it sucks. yep
|
# ? May 24, 2018 03:00 |
|
folks do we have any news about PL/I? I hear it’s an up and coming plang
|
# ? May 26, 2018 07:10 |
|
Luigi Thirty posted:folks do we have any news about PL/I? I hear it’s an up and coming plang you joke but i know at least one person from work who writes pl/x as part of their day job
|
# ? May 26, 2018 13:52 |
|
doesn't seem too bad a life really. old and clunky, but more sensible than a lot of stuff one can end up having to code in
|
# ? May 26, 2018 14:04 |
|
One of my colleagues is the about the only PL/I programmer in the country not already employed by a bank (the only people who still use PL/I), so whenever they need to perform external audits for whatever reason, he makes bank.
|
# ? May 26, 2018 15:21 |
|
carry on then posted:you joke but i know at least one person from work who writes pl/x as part of their day job please tell me pl/x is pl/i with xml syntax
|
# ? May 26, 2018 16:01 |
|
Soricidus posted:please tell me pl/x is pl/i with xml syntax nah the x means cross platform as in i5/os and z/os lol
|
# ? May 26, 2018 16:15 |
|
pl/x iss
|
# ? May 26, 2018 16:35 |
|
carry on then posted:nah the x means cross platform pl/i is already cross platform pl/x was an enhanced pl/s i have no idea what pl/s was really meant for, or how it differs from pl/i
|
# ? May 26, 2018 16:52 |
|
|
# ? Jun 4, 2024 13:13 |
|
Luigi Thirty posted:folks do we have any news about PL/I? I hear it’s an up and coming plang pl/i would be fine except that it is ibm-only it's like ibm's very own proprietary version of bcpl or c
|
# ? May 26, 2018 16:53 |