| --- Log opened Fri Sep 23 00:00:44 2016 |
| 00:27 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has quit [[NS] Quit: Leaving] |
| 00:27 | | catadroid` is now known as catadroid |
| 02:39 | <&ToxicFrog> | Hmm. There are people here who python. |
| 02:39 | <&ToxicFrog> | Is there a convenient way to turn a function that returns a different value each time it's called into a generator? |
| 02:39 | <&ToxicFrog> | Like, I have next_frame(fd), which returns the next ttyrec frame from the given fd |
| 02:40 | <&ToxicFrog> | (or None at eof) |
| 02:40 | <&ToxicFrog> | I want to iterate over all the frames |
| 02:40 | <&ToxicFrog> | In lua this is just: for frame in next_frame,fd do |
| 02:40 | <~Vornicus> | def iterize(f): |
| 02:40 | <&ToxicFrog> | In python it doesn't seem to be that simple |
| 02:40 | <&ToxicFrog> | I tried wrapping it in a while true yield, but that blows up at the end of iteration because it just yields the None |
| 02:41 | <~Vornicus> | result = f(); while result is not None: yield result; result = f() |
| 02:44 | <&ToxicFrog> | Thanks |
| 02:45 | <&ToxicFrog> | I think the sticking point is that I was expecting returning None to end the iteration |
| 02:45 | <&ToxicFrog> | And that returning nothing and returning None would be eqv |
| 02:46 | <&ToxicFrog> | And apparently neither of those are true |
| 02:48 | <~Vornicus> | returning nothing is equivalent to returning None in all situations I can think of |
| 02:49 | <~Vornicus> | to end iteration you need to raise StopIteration, technically, or reach the end of the generator function |
| 02:49 | <&ToxicFrog> | Wait |
| 02:49 | <&ToxicFrog> | So falling off the end of the function and returning nothing are not the same thing? |
| 02:50 | <~Vornicus> | okay okay so |
| 02:50 | <~Vornicus> | For regular functions, the following three are equivalent: def f(): blah blah blah; def f(): blah blah blah return; def f(): blah blah blah return None; |
| 02:52 | <~Vornicus> | For generators, you don't use *return* at all; yielding at all -- including yielding None -- waits there and continues where it left off next time you (or the loop, implicitly) calls generator.next() |
| 02:52 | <~Vornicus> | falling off the end is equivalent to raising StopIteration |
| 02:53 | <&ToxicFrog> | Aah. |
| 02:55 | <~Vornicus> | oh, you can use return. Prior to 3.3, it had to be empty, and was equivalent to raise StopIteration(); 3.3+ has return something equivalent to raiseStopIteration(something) |
| 03:21 | | * ToxicFrog befuddles |
| 03:21 | <&ToxicFrog> | I have a bunch of Command subclasses implementing the various user commands. I want to split them into their own files. |
| 03:22 | <&ToxicFrog> | So I take RegisterCommand and shuffle it off into commands/RegisterCommand.py |
| 03:22 | <&ToxicFrog> | And then call 'import commands.RegisterCommand' |
| 03:22 | <&ToxicFrog> | And it dies with "TypeError: module.__init__() takes at most 2 arguments (3 given)" |
| 03:22 | <&ToxicFrog> | And then points to this line as the cause: |
| 03:22 | <&ToxicFrog> | File "/home/ben/devel/doomrl-server/commands/RegisterCommand.py", line 3, in <module> |
| 03:22 | <&ToxicFrog> | class RegisterCommand(Command): |
| 03:23 | <&ToxicFrog> | I think I need to read more about how python modules work, because it's clearly nothing like what I'm used to. |
| 03:23 | <&Derakon> | Does RegisterCommand.py know what Command is? |
| 03:23 | <&Derakon> | Specifically, is the Command symbol a class or a module? |
| 03:24 | <&Derakon> | (See http://stackoverflow.com/questions/14583761/typeerror-module-init-takes-at-most- 2-arguments-3-given ) |
| 03:27 | <&ToxicFrog> | Oh, I think I see what's happening |
| 03:27 | <&ToxicFrog> | When I "import command.Command" to get the superclass, I actually get the module containing the superclass; I want command.Command.Command |
| 03:27 | <&ToxicFrog> | There has got to be a better way to do this |
| 03:27 | <&Derakon> | Yes. |
| 03:28 | <&Derakon> | You could add a __init__.py in the command directory that imports the Command class. |
| 03:28 | <&Derakon> | Then you could import command.Command and get the class. |
| 03:28 | <&Derakon> | You should probably rename Command.py if you do that, though. |
| 03:28 | <&Derakon> | Any symbols declared in an __init__.py...or is it __module__.py? I forget... |
| 03:28 | <&ToxicFrog> | __init__.py, I think |
| 03:29 | <&Derakon> | Anyway, any symbols declared or imported in that specially-named file will be available when you import the package. |
| 03:29 | <&Derakon> | Many libraries have a root-level __init__.py that imports a whole bunch of crap, so you can e.g. do "import numpy" and get access to numpy.zeros, numpy.float32, etc. |
| 03:30 | | * ToxicFrog nods |
| 03:32 | <&Derakon> | More generally, any module that imports X can also access anything that X imports, which is part of why "import *" is such a bad idea. I once had to trace a function across five modules, none of which imported it explicitly by name and one of which renamed it for added fun. |
| 03:32 | <&Derakon> | Like, I was searching for "func2", and one of the modules imported * from some other module, and then had a line somewhere in there saying "func2 = func". |
| 03:32 | <&ToxicFrog> | Lovely. |
| 03:33 | <&Derakon> | And that was the only reference to func in the entire module! |
| 03:33 | <&ToxicFrog> | In this case, I just need the classes to be defined; they have a metaclass that registers them in the global command table. |
| 03:33 | <&Derakon> | Ahh, reflection. |
| 03:33 | <&ToxicFrog> | Then I just need to be able to import the command table so main() can find them. |
| 03:48 | | catadroid` [catadroid@Nightstar-fk3dgo.dab.02.net] has joined #code |
| 03:50 | | catadroid [catadroid@Nightstar-eopd2k.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 05:29 | | Derakon is now known as Derakon[AFK] |
| 07:02 | | Kindamoody[zZz] is now known as Kindamoody |
| 07:34 | | catadroid` is now known as catadroid |
| 07:58 | | catadroid [catadroid@Nightstar-fk3dgo.dab.02.net] has quit [[NS] Quit: Bye] |
| 08:10 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Operation timed out] |
| 08:48 | | Pink [user1@Nightstar-g7hdo5.dyn.optonline.net] has quit [Ping timeout: 121 seconds] |
| 08:48 | | Pink [user1@Nightstar-g7hdo5.dyn.optonline.net] has joined #code |
| 08:56 | | celticminstrel [celticminst@Nightstar-f4fn3d.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 09:44 | | You're now known as TheWatcher[lab] |
| 10:27 | <@gnolam> | ... phew |
| 10:27 | <@gnolam> | Thought I would have to roll back everything I did the last couple of days there. |
| 10:30 | <@gnolam> | Crisis averted. Some of the old cargo cult code got cleaned up in the process. |
| 10:47 | <@abudhabi> | Cargo cult code? |
| 10:48 | <@abudhabi> | Did it try to imitate something that worked, but failed to understand how and why it worked, and therefore didn't? |
| 11:27 | | gizmore [kvirc@Nightstar-ofjk5v.dip0.t-ipconnect.de] has quit [Operation timed out] |
| 11:28 | | gizmore [kvirc@Nightstar-k7s4tb.dip0.t-ipconnect.de] has joined #code |
| 11:45 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 11:45 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 12:01 | | You're now known as TheWatcher |
| 12:19 | | You're now known as TheWatcher[afk] |
| 14:47 | | You're now known as TheWatcher |
| 15:03 | | catadroid [catadroid@Nightstar-ri2h4v.dab.02.net] has joined #code |
| 16:05 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 16:48 | | gizmore [kvirc@Nightstar-k7s4tb.dip0.t-ipconnect.de] has quit [Ping timeout: 121 seconds] |
| 16:48 | | catadroid` [catadroid@Nightstar-qj54qa.dab.02.net] has joined #code |
| 16:51 | | catadroid [catadroid@Nightstar-ri2h4v.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 17:51 | | catadroid` is now known as catadroid |
| 18:15 | | Derakon[AFK] is now known as Derakon |
| 18:27 | | celticminstrel [celticminst@Nightstar-f4fn3d.dsl.bell.ca] has joined #code |
| 18:27 | | mode/#code [+o celticminstrel] by ChanServ |
| 18:41 | | catadroid [catadroid@Nightstar-qj54qa.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 19:35 | <&McMartin> | abudhabi: That is indeed exactly the image intended by that phrase |
| 19:35 | <&McMartin> | Except unlike the old cults, sometimes it actually works, because the Words turn out to have power after all |
| 19:35 | <&McMartin> | When something about the environment changes, though, everything breaks and nobody knows how to fix it~ |
| 21:03 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has joined #code |
| 21:49 | <&ToxicFrog> | Hmm. |
| 21:50 | <&ToxicFrog> | I definitely dislike cone less than any other mailreader I've tried so far |
| 21:50 | <&ToxicFrog> | But it's missing some kind of important features |
| 21:50 | | Emmy [Emmy@Nightstar-9p7hb1.direct-adsl.nl] has quit [Ping timeout: 121 seconds] |
| 21:50 | <&ToxicFrog> | Like "reply to destination address", "put reply in same folder", and a search command that actually lets you look at the search results |
| 22:07 | <@celticminstrel> | Hmm... it is guaranteed that long can hold an unsigned short, right? |
| 22:08 | < catalyst> | yes |
| 22:09 | <&McMartin> | ... is that "signed long", "unsigned short"? |
| 22:09 | <@celticminstrel> | Yes. |
| 22:10 | <&McMartin> | I know short <= int <= long, but I'm not sure if short < long. |
| 22:10 | <&ToxicFrog> | McMartin: I think it's guaranteed that the signed and unsigned versions fit in the same space |
| 22:10 | <@celticminstrel> | Maybe I should use int64_t just to be safe. (Plus it'll be promoted to that regardless by the library.) |
| 22:11 | <&ToxicFrog> | It's definitely the case on ia32 and amd64 machines. |
| 22:12 | <&McMartin> | ToxicFrog: What I mean is, if sizeof(short) == sizeof(long), than an unsigned short may lose its top bit once it is interpreted as a sign |
| 22:12 | <&ToxicFrog> | Hmm. |
| 22:12 | <&McMartin> | But then, I interpreted the question as "can it hold it without distortion" |
| 22:12 | <@celticminstrel> | Yes. |
| 22:12 | <@celticminstrel> | The value must not change. |
| 22:13 | <@celticminstrel> | (Admittedly the actual numbers stored are unlikely to approach the limits, but still.) |
| 22:13 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has left #code ["Leaving"] |
| 22:13 | <&ToxicFrog> | Hmm. I think it just gets reinterpreted, not truncated, but I'm rusty on the minutiae of C casting |
| 22:14 | <&McMartin> | Yeah, I'm digging on this |
| 22:14 | <&McMartin> | catalyst's word is usually good enough and digging into it more... |
| 22:14 | <&McMartin> | ... long is guaranteed to be at least 32 bits |
| 22:14 | <&McMartin> | short is guaranteed to be at least 16 bits |
| 22:14 | <&McMartin> | It is technically compliant for short and long to both tbe 32 bits |
| 22:15 | <&McMartin> | In which case this is not the case |
| 22:15 | <&McMartin> | But if you have a short you should be sticking to the 16-bit range |
| 22:15 | <&McMartin> | That *will* fit in a long, guaranteed |
| 22:15 | <&McMartin> | If exact bit widths matter, it is The Year Of Our Lord 2016 and stdint.h should exist everywhere |
| 22:15 | <&McMartin> | And you can straight-up use uint16_t and sint32_t. |
| 22:15 | <&McMartin> | Er, just int32_t for the latter |
| 22:16 | <@celticminstrel> | I think I only used a short because I don't expect to store abnormally large numbers in it. |
| 22:16 | <&McMartin> | In all seriousness, use int, not short, that's the one that's probably register width and thus the fastest to operate on |
| 22:16 | <@celticminstrel> | And I need to cast because the library I'm using chokes on unsigned integers (though I suppose I could patch the library instead). |
| 22:23 | <@celticminstrel> | Next question: How safe is it to cast between int64_t and uint64_t? |
| 22:24 | <&McMartin> | Going from one to the other and back is lossless |
| 22:24 | <&McMartin> | You'll get garbage in the "other" if the int64_t was negative or if the uint64_t value is larger than the maximum positive representative number |
| 22:24 | <@celticminstrel> | That seems good enough for me. |
| 22:25 | <&McMartin> | But if it's in the part of their range where they overlap (0 through 2^63-1, basically), then the value is the same on both sides |
| 22:25 | <&McMartin> | If your architecture is 2's Complement (spoiler alert: it is) then you can even do math with the "garbage" values and the overflow will make it work out. |
| 22:28 | <@celticminstrel> | Okay, so given this, I think I'm just going to patch the library. |
| 22:30 | <@celticminstrel> | ...why is this library separately checking for long and int32_t? |
| 22:30 | <@celticminstrel> | And also short separately. |
| --- Log closed Sat Sep 24 00:01:00 2016 |