| timo | we could introduce sooooo many different fake codepoints in the negative number space with absurd behaviours like that ... | 00:05 | |
| ShimmerFairy | I do wonder if we should at least issue a warning when a string is constructed with a surrogate codepoint, at least in some cases. Not sure if that would be better or worse than waiting until you try to encode the text to a UTF. | 00:23 | |
| librasteve | ShimmerFairy: I think that modula’s example (naively encoding random ints) is the sort of thing that I would do (ie a reliance on non specified surrogate subtleties) | 06:37 | |
| ShimmerFairy: so I, for one, would like to see such a warning if they are stored to Str | 06:39 | ||
|
07:29
finanalyst joined
09:03
finanalyst left
10:05
finanalyst joined
|
|||
| lizmat | meh... CLIENT:: logic doesn't really work from the setting (if you expect the setting to be a single compunit, which it is) | 11:00 | |
|
11:01
finanalyst left
|
|||
| lizmat | ok, so deprecating "-".IO is simple | 13:15 | |
| at first sight | 13:16 | ||
| but does that mean that IO::Path.new("-") should also be deprecated? | |||
| and IO::Handle.new("-") | |||
| and $*ARGFILES = "-" ? | |||
| the use of "-" to indicate STDIN (or STDOUT for that matter) is pretty pervasive | 13:17 | ||
| (and open("-")) ? | |||
| fwiw, there are roast tests for $*ARGFILES handling "-" | 13:18 | ||
| ab5tract | It should just try to open a file named “-“ | 16:35 | |
| I don’t know how pervasive it is in user code | 16:36 | ||
| I would just remove any and all associations to STDIN (it can also be used to refer to STDOUT?? :O) | 16:37 | ||
| mc2 | my 2 cents: IO::CatHandle is the class where "-" should handle $*OUT | 16:58 | |
| $*ARGFILES is CatHandle, right ? | 17:00 | ||
| wow ... 6pm I have to move | |||
| timo: if you anwser I can work few hours on it tomorrow | 17:01 | ||
| (answer the mail on pkg- I mean) | |||
| lizmat | ab5tract: ok, but then how would one indicate that e.g. a program should read from STDIN ? | 17:10 | |
| timo | hi mc2, i'll take a look | 17:11 | |
| ab5tract | Isn’t that what $*IN is for? | ||
| lizmat | so I'm wondering, apart from the mention in the docs, where the decision was made to deprecate "-"'s special meaning (especially since we have tests for the use of "-") | ||
| I have a script that takes a file name to process, but instead I want it to read from STDIN | 17:12 | ||
| in the unix world, you'd specify "-" as the filename | |||
| ab5tract | Hmm | 17:13 | |
| lizmat | m: $*ARGFILES = "-"; say slurp | ||
| camelia | No such method 'slurp' for string '-'. Did you mean 'Slip'? in block <unit> at <tmp> line 1 |
||
| lizmat | m: $*ARGFILES = ("-",); say slurp | ||
| camelia | No such method 'slurp' for invocant of type 'List'. Did you mean 'Slip'? in block <unit> at <tmp> line 1 |
||
| lizmat | m: $*ARGFILES = ("-",); say $*IN.slurp | 17:14 | |
| camelia | »Wann treffen wir drei wieder zusamm?« »Um die siebente Stund‘, am Brückendamm.« »Am Mittelpfeiler.« »Ich lösche die Flamm.« »Ich mit« »Ich komme vom Norden her.« »Und ich vom Süden.… |
||
| lizmat | m: say $*IN.slurp | ||
| camelia | »Wann treffen wir drei wieder zusamm?« »Um die siebente Stund‘, am Brückendamm.« »Am Mittelpfeiler.« »Ich lösche die Flamm.« »Ich mit« »Ich komme vom Norden her.« »Und ich vom Süden.… |
||
| lizmat | hmmm | ||
| timo | mc2: if you're at the keyboard right now, we can go through the process | 17:22 | |
| uh i spoke too soon | 17:23 | ||
| but you can see from the README.source.org that we use the gbp (git-buildpackage) suite of tools to handle everything from getting the newest version (uscan), keeping the actual upstream stuff tracked in git (pristine-tar), and some part of it does the necessary patching out of stuff (i think the build package -s step is what does all that?) | 17:25 | ||
| debian-mentors recommends building branches debian/releasename-backports and work in there | 17:28 | ||
| dep-team.pages.debian.net/deps/dep14/ this is relevant | 17:44 | ||
| ab5tract | The “-“.IO story is indeed more complicated than I thought at first sneeze :( | 19:24 | |
| lizmat | it even gets a lot harder to provide a good deprecation message for 2 reasons: | 19:25 | |
| .IO is called from within the core in many places, so the caller's language level is not the language level of the actual code that is doing a call to some functionality | 19:26 | ||
| like open("-") | 19:27 | ||
| so it may deprecate while the users code is 6.c, or not die if the user's code is 6.e | |||
| and then there's the issue of producing the correct file / line | 19:28 | ||
| and finally there's the question: why do we want to remove this feature at all, as it is so deeply interwoven in the core | 19:29 | ||
| going afk to frobnicate on that& | |||
| mc2 | timo: ok ... so I create a trixie-backport branch and try to rebuild all the raku packages ? | 20:37 | |
| in unix, - is $*IN or $*OUT depending of the context | 20:38 | ||
| tcc -o - - means compile the C program from stdin to stdout | 20:39 | ||
| this must be handled by the scripter, not raku ... unless for the parts that are definitely unix culture related ( CatHandler and $*ARGFILES are the 2 I have in mind ) | 20:42 | ||
| ShimmerFairy | It should be noted that, in POSIX, '-' is not universally treated as stdin/stdout. Sometimes it just means a file named '-'. So actually, IIUC, the current behavior makes it impossible to work with a file literally named '-' like some POSIX utils can, which seems bad. | 21:59 | |
|
22:59
kjp left
23:00
kjp joined
23:01
kjp left
23:02
kjp joined
|
|||