search of the:   for: 
code logs -> 2016 -> Wed, 08 Jun 2016< code.20160607.log - code.20160609.log >
--- Log opened Wed Jun 08 00:00:32 2016
00:04 Derakon[AFK] is now known as Derakon
00:05 gnolam [quassel@Nightstar-t2vo1j.tbcn.telia.com] has quit [[NS] Quit: Z?]
00:20 Reiv [NSwebIRC@Nightstar-q8avec.kinect.net.nz] has quit [Ping timeout: 121 seconds]
00:22 Reiv [NSwebIRC@Nightstar-q8avec.kinect.net.nz] has joined #code
00:22 mode/#code [+o Reiv] by ChanServ
00:30 Turaiel[Offline] is now known as Turaiel
00:42 Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection reset by peer]
00:52 Shady [ShadyGuru@Nightstar-ut3scg.t132.ptd.net] has joined #code
01:43 thalass [thalass@Nightstar-283.o7s.158.104.IP] has joined #code
01:43 mode/#code [+o thalass] by ChanServ
05:06 Derakon is now known as Derakon[AFK]
05:10 catadroid` [catadroid@Nightstar-ugedcl.dab.02.net] has joined #code
05:12 catadroid [catadroid@Nightstar-s5gesh.dab.02.net] has quit [Ping timeout: 121 seconds]
05:18 Turaiel is now known as Turaiel[Offline]
05:31
< Shady>
Hello. I currently use HexChat, and I wanna recommend it to someone else. Does anyone have the download link?
05:34 Reiv [NSwebIRC@Nightstar-q8avec.kinect.net.nz] has quit [Ping timeout: 121 seconds]
05:57 * Shady forgot he posted here, used google already, found it
05:58
<&McMartin>
Hooray
06:07
< Shady>
Almost have her here :P
06:07
< Shady>
Well, a different channel
06:15 * catadroid` spins
06:15 catadroid` is now known as catadroid
06:15
< Shady>
Wow, massive change, I don't even recognize you now o.o
06:15 Alek [Alek@Nightstar-9qtiqv.il.comcast.net] has quit [Ping timeout: 121 seconds]
06:16
<@celticminstrel>
XD
06:18
< catadroid>
...
06:19 catadroid is now known as elayne
06:54
<@abudhabi>
You can obtain electricity with an appropriate device and putting it next to a power line, by induction, yes?
06:57
< elayne>
What's your base case?
06:57
<@ErikMesoy>
This sounds like the start of a bad idea.
06:58
<@abudhabi>
My coworkers were talking about somehow making a drone recharge from landing on power lines.
06:59
<@abudhabi>
I argued that it probably would involve very heavy transformers for the high voltage lines, but I'm not 100% sure I'm right.
07:01 Crossfire [Z@Nightstar-r9lk5l.cust.comxnet.dk] has quit [Ping timeout: 121 seconds]
07:01
<@ErikMesoy>
Sounds like a slightly less bad idea, but definitely possible if you throw enough engineers at it.
07:36 Shady [ShadyGuru@Nightstar-ut3scg.t132.ptd.net] has quit [[NS] Quit: Yay, he's gone]
07:46 elayne [catadroid@Nightstar-ugedcl.dab.02.net] has quit [[NS] Quit: Bye]
10:30
<@ErikMesoy>
Is there a particular encouraged way to store a lot of boolean flags in Python? It looks a little odd to just declare UserProfile.flag1=True, UserProfile.flag2=True etc. as I'm doing now.
10:33
<@pjdelport>
What's the context?
10:33
<@pjdelport>
That should be fine for your average thing, unless you need something like a bitmap instead?
10:33
<@abudhabi>
Why not use a collection of some sort?
10:40
<@TheWatcher>
bitwise operators time!~
10:40
<@ErikMesoy>
Context is measuring stuff user has done/set/found/unlocked/toggled, and intuitive collections seemed like pointless overhead.
10:41
<@ErikMesoy>
Bitwise operators sound like a great way of confusing code maintainer down the line (probably also me).
10:42
<@pjdelport>
ErikMesoy: Do you want a dict or set, perhaps?
10:43
<@ErikMesoy>
pjdelport: Maybe what I should be asking is "what do dicts or sets get me here?" other than overhead?
10:43
<@ErikMesoy>
Besides, python. Everything is a dict anyway. _Ã
10:44
<@pjdelport>
Well, you probably shouldn't worry about overhead unless you're optimising for something and benchmarking. Intuition can lead you astray. :)
10:44
<@pjdelport>
Sets can be more memory-compact, if that's what you're worried about.
10:44
<@pjdelport>
They avoid having both keys and values.
10:44 Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code
10:44 mode/#code [+qo Vornicus Vornicus] by ChanServ
10:44
<@ErikMesoy>
I mean coding overhead. Performance be damned.
10:45
<@pjdelport>
Oh, then yeah, same thing. Sets can lead to very compact code.
10:45
<@pjdelport>
(I love Python's sets, and wish more languages had them as natively.)
10:45
<@ErikMesoy>
Hmm. sets.
10:45
<@pjdelport>
It depends entirely on what you're actually doing, though.
10:46
<@pjdelport>
How does the code use it, and what kind of operations do you need on the flags?
10:46
<~Vornicus>
<3 sets
10:47
<@ErikMesoy>
I need operations of set flag, clear flag, check flag. This is supposed to be a load of booleans for e.g. first time info popups, user toggles.
10:47
<@ErikMesoy>
if flag, display help, else do nothing.
10:48
<@ErikMesoy>
If verbose flag, display long message, else display short message.
10:50
<@pjdelport>
Command line flags?
10:50
<@pjdelport>
Shouldn't you just be using argparse or something rather than maintaining your own flags for that?
10:51
<@ErikMesoy>
No.
10:51
<@ErikMesoy>
Well, some of these could theoretically be passed by command line, but mostly they're supposed to be user-modifiable from within the program
10:52
<@ErikMesoy>
User should be able to toggle "verbose mode" without having to restart the program and give a new argument at the command line.
10:53
<@ErikMesoy>
Maybe "show/hide tooltips" is a better comparison. Checkbox stuff.
10:53
<@pjdelport>
Ah, so settings?
10:53
<@pjdelport>
You probably want to use a settings backend of some sort.
10:53
<@pjdelport>
There's ConfigParser built in, but it's not the most convenient API.
10:54
<@ErikMesoy>
Right. I think I'll maybe stick with raw object properties for now and come back and ask again when I have more use cases and a more pressing need to drive the refactor tractor and learn a backend, because this is getting a little out there for what's at present just one setting and me trying to future-proof.
10:55
<@pjdelport>
There are a bunch of projects that give you more convenient APIs, and should avoid the need to manually manage attributes like that.
10:55
<@ErikMesoy>
But future-proofing may be spilling into premature optimization here.
10:55
<@pjdelport>
You should be able to declare your schema, and then just access flags through the appropriate interface.
11:27
<@TheWatcher>
Horay, I now need to use 4 dimensions for this diagram.
11:27
<@TheWatcher>
(killmenow)
11:35
<@abudhabi>
TheWatcher: https://i.warosu.org/data/tg/img/0459/44/1457759085223.jpg
11:40
<@pjdelport>
Hah, awesome
12:12
<@TheWatcher>
ACtually, more https://www.youtube.com/watch?v=FnbYcB9ctu8
12:26
<@ErikMesoy>
I just want to say that anonymous lambda functions are not only magic, they are magic in ways it would have been hard for me to predict I would have needed or wanted in the way I am currently using some.
12:27
<@abudhabi>
Aren't lambdas in Python extremely limited?
12:28
<@abudhabi>
I mean, they're only one statement.
12:29
<~Vornicus>
Sure, but that's often enough
12:29
<@ErikMesoy>
That's kinda the point here - they are *small* and can be inlined.
12:31
<@pjdelport>
abudhabi: lambdas and defs are just different syntax for the same thing, in Python.
12:31
<@ErikMesoy>
Right now I'm using them as dict values so that I can do "when given key X, run function corresponding to X" and that function does not need to be defined anywhere else.
12:32
<@ErikMesoy>
The dict basically goes {x: lambda: function(argument)}
12:32
<@pjdelport>
So it's more a lexical question than one of functionality. If you need statement syntax, you need a def; if you only need expression syntax, you can use a lambda.
12:32
<@pjdelport>
The def style is generally encouraged for readability, which makes sense in Python.
12:32
<@pjdelport>
lambdas tend to be more obscuring in that paradigm.
12:33
<@pjdelport>
(And I say that as a Haskeller, where lambdas are ubiquitous, so I'm not biased against them.)
12:37
<@abudhabi>
ErikMesoy: Ah, you're storing functions in a map. That's pretty cool.
12:37
<@ErikMesoy>
abudhabi: And now you have me thinking how to sneak in more than one statement.
12:40
<@pjdelport>
You can wrap the usual statement syntax things into defs, and then just build up expressions with that.
12:40
<~Vornicus>
storing functions in a map is amazingly useful.
12:40
<@pjdelport>
But that's the kind of thing you see in obfuscated Python code contests, not in real code (hopefully :).
12:47 catadroid [catadroid@Nightstar-r7d8vk.dab.02.net] has joined #code
13:02
<@TheWatcher>
pjdelport: I have to admire your optimism~
13:08
< catadroid>
My colleague appears to have beaten the msvc compiler into deducing a reference as a pointer
13:08
< catadroid>
...I'm not going to play compiler golf with this, I'm not going to play compiler golf with this...
14:07
<@TheWatcher>
Yes you are~
14:23
<@abudhabi>
Does "object" have any special meaning in Python?
14:24
<@pjdelport>
abudhabi: It's the default base type.
14:26
<@abudhabi>
OK. Need to open a thesaurus. :V
14:27
<@pjdelport>
Basically, "class foo: ..." is short for "class foo(object): ..."
14:27
<@pjdelport>
Normal classes are all subclasses of object.
14:49
< catadroid>
I'm averaging four lines of commenting per loc
14:49
< catadroid>
Optimising containers, everybody
14:51 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has joined #code
14:51 mode/#code [+o Emmy] by ChanServ
15:06 Crossfire [Z@Nightstar-r9lk5l.cust.comxnet.dk] has joined #code
15:06 mode/#code [+o Crossfire] by ChanServ
15:13
<@simon_>
someone here linked to a blog post about some Rust type system problems some weeks ago. was it Vornicus, jerith, McMartin?
15:13
<~Vornicus>
I think it was mcm; he was going on about traits. Let me search the second sight
15:14
<@simon_>
Vornicus, awesome. yeah, I think it was mcm, too.
15:17
<~Vornicus>
no luck. :(
15:19
<@simon_>
nah, me neither. not even on hn.algolia.com (hacker news search engine)
15:20
<@TheWatcher>
https://starforge.co.uk/irclogs/code/code.20160526.log.html#m2108-0
15:21
<@TheWatcher>
No blog post. though
15:21
<@TheWatcher>
Not sure if that's what you meant.
15:21 * TheWatcher continues hunting
15:22
<@simon_>
TheWatcher, that's not it, but thanks for looking. it's ok, I'll probably run into the blog post again.
15:22 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds]
15:30 * TheWatcher eyes his new schema diagram
15:32 Derakon_ [chriswei@Nightstar-5mvs4e.ca.comcast.net] has joined #code
15:34 Derakon[AFK] [chriswei@Nightstar-5mvs4e.ca.comcast.net] has quit [Ping timeout: 121 seconds]
15:36
<&ToxicFrog>
simon_: possibly http://smallcultfollowing.com/babysteps/blog/2016/04/27/non-lexical-lifetimes-in troduction/ ?
15:50
<@simon_>
ToxicFrog, exactly! thanks!
17:07 Shady [ShadyGuru@Nightstar-ut3scg.t132.ptd.net] has joined #code
17:07 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has joined #code
17:07 mode/#code [+o Emmy] by ChanServ
17:08 crystalclaw|AFK is now known as crystalclaw
17:15
< catadroid>
Sometimes I feel like a wizard
17:15
<@Azash>
Hmm?
17:16
< catadroid>
I'm implementing bit twiddle operations to make this hash map as fast as possible
17:16 * Vornicus wizard hats catadroid
17:16
<@Azash>
Vornicus: Don't try to robe catadroid into anything untoward now
17:17 * Vornicus is not bloodninja.
17:30 * ErikMesoy cracks knuckles, begins migration from Python 2 to 3.
17:30 * Azash mutters "at least it's not rails 2 to 3"
17:31
<@Azash>
A friend of mine often says about coding work that everyone has their personal vietnam experience
17:32
<@Azash>
For me it was spending about 15 months with the occasional week or two doing other stuff on migrating a rails project, written to be heavily coupled into rails and active<component> internals, from early 2 to late-ish 3
17:32
<@Azash>
Alone, confused and with no documentation
17:32
<@Tamber>
"You weren't there, man. You weren't there."
17:37
<&McMartin>
4 months now that I haven't been able to get a produced binary from Xcode that meets my standards
17:37
<&McMartin>
Though I've learned a few new tricks
17:37
<&McMartin>
Things that were less obnoxious: that time I was working on debugging a crash that only happened if you ran as administrator
17:38
<@TheWatcher>
... wat
17:39
<&McMartin>
There was some trampoline code being generated at build time and then being injected into place...
17:39
<&McMartin>
... and it was using a 32-bit relative jump for a call.
17:39
<&McMartin>
A call in a different DLL
17:39
<&McMartin>
A DLL subject to ASLR when run as administrator.
17:39
<&McMartin>
Thus putting it more than 2GB away
17:39
<@TheWatcher>
>.<
17:39
<&McMartin>
We had to replace the generated trampoline code with assembly code with a 64-bit absolute jump.
17:41
<@TheWatcher>
That must've been great fun to find!
17:41
< catadroid>
...I can now intuitively implement move constructors
17:41
< catadroid>
Am I a monster?
17:42
<@Azash>
Tamber: It was a Project, really
17:42
<@Azash>
My favourite part was the bespoke RBAC
17:43
<@Azash>
Consisting of a long string literal with interpreted variables in it that was then passed to ruby's method creating method
17:43
<@Tamber>
;_;
17:45
<@Azash>
Tamber: Also things like activerecord_base_extensions.rb, or increase_rack_key_space_limit.rb
17:45
<@Azash>
Oh oh let me tell you about the second one
17:45
<@Azash>
Basically it also had a bespoke JS frontend framework
17:45
<@Azash>
yes, a bespoke framework
17:45
<@Azash>
Which among other things did not do things the REST way
17:45 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds]
17:45
<&McMartin>
catadroid: I've probably spent too much time staring into the abyss myself to map that...
17:46
<&McMartin>
... but isn't a move constructor a copy constructor that lets you trash the place on the way out? Is there some other trick?
17:46
<@Azash>
Rather, CRU was done by passing around the entire current page/document, in essentially serialized JSON format
17:46
<@Azash>
Which eventually caused the problem that some pages featured too many objects and ended up exceeding Rails' POST body size limit
17:47
<@Azash>
Not size as in actual size
17:47
<@Azash>
Size as in the amount of keys in a single nested JSON structure that Rack would accept
17:48
<@Azash>
The last fix upgraded the Rack limit to a 1,048,576 key JSON structure
17:48
<@Azash>
It's been stable for a while
17:48
<@Azash>
So here's hoping
17:50
<@ErikMesoy>
Hmm. Error at "import wx", which worked for Py2 and not for Py3. That's a nonobvious place to err.
17:53
<@ErikMesoy>
wxwidgets for python apparently not fully ported to 3 yet, needs Wx Project Phoenix. I sense yak shaving ahead...
17:56
< catadroid>
A move constructor lets you trash the input instance, although that instance needs to be valid to destruct
17:57
<&McMartin>
... so, you can trash the place but you can't take out any load-bearing walls. >.>
17:57
< catadroid>
Which means you can steal the internal representation instead of needing to copy it
17:57
< catadroid>
Heh, yeah
17:57
<@ErikMesoy>
...wxPythonPhoenix's installer page tells me how to install it for/with Python *2*!
18:01
<&jerith>
ErikMesoy: Why are you proting from 2 to 3?
18:01
<&jerith>
*porting
18:02
<&[R]>
Because Python 2 is deprecated?
18:02
<@TheWatcher>
Tell that to lots of library maintainers~
18:03
<@ErikMesoy>
jerith: Maybe I'm not, after all. >_> But I figured I should, would have to eventually, and I was gradually seeing more Py3 results on Stack Exchange when I went to look shit up.
18:03
<&jerith>
But Python 2.7 will still be supported for a very long time.
18:03
<&jerith>
For a lot of existing stuff, porting really isn't worth the effort.
18:03
<&jerith>
For new stuff, py3 probably makes sense.
18:04
<&jerith>
For libraries, supporting py2 and p3 makes sense.
18:04
<&jerith>
If you want decent performance, stick with py2 and use pypy. :-)
18:04
<@ErikMesoy>
What does it mean for me that Python 2.7 "will be supported"?
18:04
<@ErikMesoy>
I haven
18:04
<&jerith>
Security fixes, mostly.
18:05
<&jerith>
And the interpreter isn't going to go away or anything.
18:05
<@ErikMesoy>
I haven't ever contacted anything like official support. And unofficial support, as mentioned, is me seeing answers for Py3 popping up.
18:06
<&jerith>
I meant "py2 isn't being abandoned".
18:06
<&jerith>
It can't be, there's just too much existing code that will never be rewritten in py3.
18:10 catadroid` [catadroid@Nightstar-co9dqb.dab.02.net] has joined #code
18:11 * ErikMesoy shakes head
18:12
<@ErikMesoy>
"Just because Py3 is eight years old doesn't mean you should update. Just because wxPython for Py3 is two years old doesn't mean it's anywhere near ready."
18:12
<@ErikMesoy>
This feels really strange.
18:13 catadroid [catadroid@Nightstar-r7d8vk.dab.02.net] has quit [Ping timeout: 121 seconds]
18:16
<&jerith>
The thing about py3 is that it's basically a different (but very similar) language.
18:17
<&jerith>
It's *really* hard to write C extension modules that support both without a lot of ugly #ifdefery.
18:17
<@ErikMesoy>
Very very similar indeed. 2to3 says my current project needed no changes.
18:18
<&jerith>
(Which is why all sensible people use cffi to interface with C code these days.)
18:18
<@ErikMesoy>
No changes in *my* code at least. Seems a lot of changes required in what I'm importing though. >_>
18:19
<&jerith>
2to3 only knows about syntax and stdlib.
18:20 * McMartin eyes his email
18:20
<&McMartin>
Apparently there was action on a bug I posted to Apache nine years ago
18:25
<&jerith>
It's generally not that hard to write new code that is compatible with both py2 and py3.
18:26
<@ErikMesoy>
But here I hadn't been trying to do that. It just happened. *shrug*
18:27
<~Vornicus>
mcm: man
18:34
<&[R]>
McMartin: at least they didn't do what the Quassel guys did
18:34
<&[R]>
Which was close every bug that was over a year old, regardless of wether or not it was fixed.
18:35
<&[R]>
Bugs and feature requests.
18:35 * [R] got 13 emails from their tracker all at once as a result
18:36
<&McMartin>
Yeah, actually, I think this one may have been fixed back then, and some confused person added a pull request to it
18:39 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has joined #code
18:39 mode/#code [+o Emmy] by ChanServ
18:43
<&jerith>
I had a bug I reported (and provided a patch for) fixed about six years later.
18:43
<&jerith>
Bugs I fixed in Python also took multiple years to be merged.
18:49
<&jerith>
(I do not have a very high opinion of the core python devs.)
19:16 Emmy [M@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds]
19:48
<@TheWatcher>
[R]: ... wat
19:49
<@Tamber>
That's the CADT model of stopping the bug-tracker looking terrible, I think.
19:50
<@TheWatcher>
Can't be Arsed Doing This?
19:51
<&McMartin>
Cascade of Attention-Deficit Teenagers.
19:51
<@pjdelport>
https://www.jwz.org/doc/cadt.html
19:52
<@TheWatcher>
I'd laugh, but... yeah
19:54
<&[R]>
From what I've heard, it's not any better with priorietary shit
19:54
<&[R]>
Eitherway though, it's not excusable
19:58
<&[R]>
http://bugs.quassel-irc.org/issues/788#change-2550
19:59 Netsplit Krikkit.Nightstar.Net <-> Deepthought.Nightstar.Net quits: @ErikMesoy, @Ogredude, @Vornicus, @macdjord, @celticminstrel, ricer, @^Xires, @Azash, @[R], @Turaiel[Offline], (+2 more, use /NETSPLIT to show all of them)
20:00 Netsplit over, joins: @Turaiel[Offline], @Azash, @^Xires, @macdjord, ~Vornicus, @pjdelport, @ErikMesoy, @celticminstrel, ricer, @crystalclaw (+2 more)
20:01
<@ErikMesoy>
ooaoooaooooaoaoaoooooaoooooooo!
20:02
<&McMartin>
There's "excusable" and then there's "explainable", though
20:02
<@ErikMesoy>
The mode assignments remind me of https://i.imgur.com/l8fiWQ6.jpg
20:02
<&McMartin>
If you hit your truck factor - and with high enough turnover, you always will - it's probably less work to do the rewrite
20:03
<&McMartin>
But if they had component/systemic tests, keep them
20:04
< catadroid`>
It saddened me how few unit tests we had until I started writing a lot
20:14
<&jerith>
I'm trying to write fewer tests these days.
20:14
<&jerith>
(But better ones.)
20:30 Alek [Alek@Nightstar-9qtiqv.il.comcast.net] has joined #code
20:30 mode/#code [+o Alek] by ChanServ
20:31
<@Alek>
yo, visit GoG Galaxy to get System Shock 2 free.
21:03 Shady is now known as Shady|afk
21:05 ion [Owner@Nightstar-6grqph.vs.shawcable.net] has quit [Ping timeout: 121 seconds]
21:05 ion [Owner@Nightstar-6grqph.vs.shawcable.net] has joined #code
21:05 mode/#code [+o ion] by ChanServ
22:05 Reiv [NSwebIRC@Nightstar-q8avec.kinect.net.nz] has joined #code
22:05 mode/#code [+o Reiv] by ChanServ
22:39 crystalclaw is now known as crystalclaw|AFK
23:21 catadroid` is now known as catadroid
23:46 Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection reset by peer]
--- Log closed Thu Jun 09 00:00:25 2016
code logs -> 2016 -> Wed, 08 Jun 2016< code.20160607.log - code.20160609.log >
 search of the:   for: 

[ Latest log file ]