| <<<Back 1 day (to 2016/12/12) | 20161213 |
Robin_Watts | tor8, sebras: For the logs, 2 commits on robin/master | 00:55.14 |
| Morning tor8. 2 commits on robin/master | 10:12.39 |
tor8 | Robin_Watts: morning. API comment and jbig2dec update LGTM. | 10:59.34 |
Robin_Watts | tor8: Ta. | 11:33.04 |
| 2 more commits. This leaves us with the 4 const warnings in mujs, and a 'strdup' warning in harfbuzz. | 11:51.21 |
| Just gonna check on unix now. | 11:51.30 |
| ugh. lots of warnings in unix, but all look bogus at first glance. | 11:56.49 |
tor8 | huh. I keep the build warning free on linux with both gcc and clang (but not in release mode with gcc because it gets massively confused with setjmp) | 11:57.46 |
Robin_Watts | tor8: This was release mode. | 11:58.19 |
| We can never have a path which doesn't have a move as it's first command, right? | 11:58.47 |
tor8 | correct. all paths must start with move, and I think that's enforced somewhere | 11:59.26 |
Robin_Watts | tor8: So we might as well not bother storing the first command. | 11:59.48 |
tor8 | fz_warn(ctx, "lineto with no current point"); | 11:59.50 |
| then we'd still need to store a flag for whether we've seen a moveto | 12:00.33 |
Robin_Watts | tor8: No, cos a cmd length of 0, and a coord length of > 0 would mean we'd seen a moveto. | 12:01.29 |
tor8 | oh, right | 12:01.35 |
| I'm not sure whether the extra mental complexity of having an implicit first moveto is worth the very minor memory gain | 12:02.09 |
| or was there some other reason? | 12:02.16 |
Robin_Watts | tor8: Actually, we can have a rectto as the first element. | 12:06.18 |
| I was thinking that it might let me get around 4 of the warnings. | 12:06.33 |
| but I don't think it's worth it. | 12:06.40 |
tor8 | ah, right. because that's a compressed moveto/3x lineto | 12:06.42 |
Robin_Watts | yeah. | 12:06.47 |
tor8 | which warnings? | 12:06.54 |
Robin_Watts | The ones about x and y being maybe used uninitialised in the path stuff. | 12:07.28 |
tor8 | yeah... those are gcc being braindead though... | 12:07.55 |
Robin_Watts | No, gcc is right, cos it can't know about the restrictions we have on the ordering. | 12:08.17 |
| In debug mode I get 3 freetype "statics defined but not used" warnings. | 12:08.44 |
| and that's all. | 12:08.48 |
| so I'm happy. | 12:08.50 |
tor8 | clang gets it right though. | 12:08.52 |
| Robin_Watts: just initialize x and y to zero if the warning bothers you | 12:09.05 |
Robin_Watts | That would bother me more :) | 12:09.14 |
| And I'd say that clang misses it, rather than gets it right, but ho hum. | 12:10.31 |
jhabjan | hi, quick question, I'm using "gsapi_run_string" to push this: (E:/test_data/a?a.pdf) (r) file runpdfbegin | 13:51.34 |
| the question mark in the filename actually represents some Greek letter | 13:52.06 |
| and I'm getting e_ioerror = -12 | 13:52.45 |
| I've used gsapi_set_arg_encoding(_gs_instance, GS_ARG_ENCODING.UTF8) to set UTF8 encoding | 13:53.24 |
| and converted all my strings to UTF8 | 13:53.31 |
| any suggestions? | 13:53.39 |
kens | rename the file ? :-) | 13:56.08 |
| I think the problem 'may be' that you are using PostScript to run the file, I'm not certain that the PostScript file operators can deal with Unicode strings. | 13:57.03 |
| I would have to go check | 13:57.08 |
jhabjan | I thought of that too... but... that would result with making another copy of the file which I'm not sure it's a good idea. | 13:58.16 |
kens | I'm reasonably sure the PostScript specification doesn't say anything about it being able to read Unicode filenames, its a bit old for that. I don't really have the time to go debugging it just now, I'm busy with a different problem. | 14:00.13 |
jhabjan | ok, no worries, thanks | 14:00.37 |
| Forward 1 day (to 2016/12/14)>>> | |