| | <<<Back 1 day (to 2018/08/28) | 20180829 |
| inflex | What is the primary unit of measurement used in muPDF for things like bbox and the position of elements? | 10:38.42 |
| | (thou?) | 10:38.45 |
| tor8 | inflex: points. 1/72 of an inch. | 10:39.09 |
| inflex | Ah, thanks, that was my 2nd option, thank you | 10:39.48 |
| avih | tor8: re mujs, i'm trying to run test262, and it seems mujs returns undefined on Object.getOwnPropertyDescriptor(Date.prototype, "getYear") | 10:46.13 |
| | (for now i'm using mpv as a harness host, as it has some file utils etc) | 10:47.24 |
| | tor8: can you confirm the behavior and the fact that it's an issue? | 10:48.18 |
| tor8 | avih: Date.prototype.getYear is NOT part of the standard. | 10:49.25 |
| | You should use Date.prototype.getFullYear. | 10:49.35 |
| | getYear suffers from the Y2K problem | 10:49.49 |
| avih | also, it seems it expects "escape" and "unescape", but it seems mujs doesn't have them. would mapping to {en,de}codeURIComponent suffice? | 10:49.56 |
| tor8 | and is not part of the ES5 spec | 10:50.03 |
| avih | huh | 10:50.27 |
| | and [un]escape ? | 10:50.35 |
| tor8 | avih: these functions are mentioned in Annex B (the chapter with the old crap that we no longer desire) | 10:50.44 |
| avih | iirc it mentions it'd be deprecated | 10:50.48 |
| tor8 | they were removed in ES5 | 10:51.00 |
| avih | yes, these tests are of annex B indeed | 10:51.05 |
| tor8 | they're trivial to add in by just writing some simple JS functions, if you really want them | 10:51.28 |
| avih | should i just ignore annexB for now? | 10:51.28 |
| tor8 | all except octal literals | 10:51.37 |
| | I would ignore annexB forever | 10:51.46 |
| | avih: http://ix.io/1lve with this script you can run many of the tests | 10:52.43 |
| avih | k. it happens to be the first suite, so i stopped to look at the failures (i'm also developing the harness and uses these tests to test the harness). maybe i should just skip annex B for now till i'm sure the harness is solid | 10:53.08 |
| tor8 | mujs explicitly does not support Annex B (deprecated functions that were removed in ES5) | 10:53.42 |
| avih | readline is available at the mujs REPL? (never actually used it for something) | 10:54.16 |
| tor8 | avih: yes, part of the 'main.c' mujs shell | 10:54.40 |
| avih | tor8: does it really work with default and uncancelable strict mode? | 10:55.00 |
| tor8 | gc, load, print, write, read, readline, quit and require are non-standard functions in the mujs shell | 10:55.08 |
| | and you need to turn off strict mode. | 10:55.16 |
| | the test262 scripts assume non-strict mode | 10:55.22 |
| avih | right. REPL should really default to strict mode off. that's the standard iirc | 10:55.35 |
| tor8 | yeah. I'm thinking of changing it to that. | 10:55.44 |
| avih | yeah. i think that's what people expect. | 10:56.39 |
| tor8 | avih: pushed. use '-s' to start in strict mode. | 11:03.06 |
| avih | nice | 11:03.13 |
| | ah, ok, i see that load is protected. | 11:04.40 |
| | (in your mini harness). | 11:04.56 |
| | how do you run it? find . | mujs <script>.js ? | 11:05.18 |
| tor8 | yes. | 11:06.26 |
| | find test262/test/suite -name '*.js' | mujs harness.js | 11:06.47 |
| avih | yeah | 11:06.53 |
| | tor8: did it find issues you were not aware of? what kinds of tests is this harness not able to drive? | 11:08.36 |
| | the harness is missing aliases $PRINT=print and $FAIL=$ERROR | 11:29.25 |
| | tor8: many test pass, but also many tests fail, and then it hangs on test/suite/ch15/15.4/15.4.4/15.4.4.18/15.4.4.18-3-14.js | 11:36.25 |
| | did you look into the failures? | 11:36.30 |
| | tests* | 11:36.35 |
| paulgardiner | I'm playing with mupdf-gl. I can alter checkboxes within forms, but I don't seem to be able to alter text widgets. | 11:47.43 |
| tor8 | avih: no, I haven't had time | 11:54.29 |
| | paulgardiner: yeah, you need to (currently) edit text widgets with the annotation editor | 11:54.47 |
| | I'm working on that | 11:54.54 |
| avih | tor8: i found all the hanging/crashing tests and skipping them allows to complete the suit | 11:54.56 |
| | (about 20) | 11:55.09 |
| paulgardiner | Is there currently a mupdf app that does in-place widget filling? | 11:56.48 |
| tor8 | paulgardiner: not at the moment, no | 11:56.58 |
| paulgardiner | Oh okay | 11:57.07 |
| tor8 | paulgardiner: but if you hit 'a', select the text widget, you can type in the value in the Value field on the right | 11:58.36 |
| paulgardiner | I was just looking for the in-place filling. I though if you'd had that working, I might of tried to do it also in muso. | 12:00.00 |
| avih | tor8: that's my updated skipping harness https://0x0.st/stMk.txt and this is its output: https://0x0.st/stMd.txt | 12:00.19 |
| | sec, let me count how many pass/fail | 12:00.43 |
| | Total: 11701 | 12:05.24 |
| | Pass: 7275 | 12:05.25 |
| | Failed: 4426 | 12:05.25 |
| | (there could be more missing harness functions, i didn't look at the entire output line by line) | 12:06.20 |
| tor8 | paulgardiner: yeah... I've been meaning to get that working, but I keep getting distracted :) | 12:51.39 |
| sebras | tor8: that distraction would be me. ;) | 12:57.38 |
| moolc | tor8: f85a9d6a08ebba9e319abdc05eadc3e443b878f1 commit comment looks truncated, almost as if someone forced you to hit enter mid-thought | 13:29.47 |
| tor8 | moolc: I blame sebras... | 14:49.20 |
| sebras | tor8: : moolc: yes, that one's on me. I noticed a minute after pushing to master. :-( | 14:51.03 |
| | moolc: it was meant to say "This commit adds the check.", so it's not that important really. | 14:51.54 |
| avih | tor8: when running each test in its own new mujs instance, 9733 pass and 1992 fail. i also got it to print its intended error message. at least most of the time: https://0x0.st/stSi.txt this is a more digestable output | 16:22.20 |
| | (the skipped tests are still skipped) | 16:23.16 |
| | i think there are some low hanging fruits there, like few expected exception "intointeger". and actual exception "TypeError: string function called on null or undefined" | 16:26.15 |
| | (some of the tests, though i think few, make some properties of the global unconfigurable, and later tests use the same properties and assume a clean slate) | 16:27.30 |
| | (hence new mujs instances) | 16:27.49 |
| | also, to get the proper messages i had to catch the exception which load() doesn't expose. so i had to add loadRaw to main.c: js_loadstring(J, js_tostring(J, 1), js_tostring(J, 2)); | 16:30.11 |
| | sorted: https://0x0.st/stST.txt | 16:43.27 |
| | this way classes of issues pop out more clearly. | 16:44.06 |
| | Forward 1 day (to 2018/08/30)>>> | |
Log of #mupdf at irc.freenode.net.