| <<<Back 1 day (to 2016/08/21) | 20160822 |
tor8 | sebras: for Array.put we already throw in the ... oh, no we don't, we just print warnings and fail silently :/ | 10:24.31 |
| soft masks can be drawn in two modes -- alpha or luminescence | 10:26.19 |
| in alpha mode, we throw away the color and keep only the alpha | 10:26.30 |
| in luminescence mode we draw and convert the final grayscale to alpha | 10:26.42 |
sebras | tor8: you around? | 11:39.57 |
tor8 | sebras: hi. | 11:48.15 |
| some comments in the logs for you | 11:48.23 |
sebras | tor8: I did. still have questions. :) | 11:49.20 |
| do you prefer me to change the underlying fz_array_put() and friends to throw if you try to put into the wrong type of objecT? | 11:50.02 |
| I feel that it is risky to make that change. | 11:50.14 |
| mainly because I have no idea how this may cause things to blow up unexpectedly. | 11:50.37 |
tor8 | I believe the decision to be that lenient was to reduce the amount of error checking required, back before we had exceptions | 11:51.04 |
sebras | indeed. | 11:51.20 |
tor8 | and then later when we didn't have a fz_context threaded through the pdf_obj functions we couldn't throw | 11:51.21 |
sebras | now we can, but do we want to? :) | 11:51.36 |
tor8 | so I think we should be okay with adding strict index checks, etc, now | 11:51.39 |
| i.e. turn all the warnings into actual exceptions | 11:51.52 |
sebras | ok. then I'll move most of these checks down from the JNI layer to the core functions. | 11:52.12 |
| tor8: how do you feel about the two commits on sebras/master? | 11:52.39 |
tor8 | sebras: typo fix LGTM | 11:57.56 |
sebras | tor8: in pdf_array_put() we return (with a comment "Can't warn :(") if obj >= PDF_OBJ__LIMIT. | 11:58.45 |
| tor8: while in pdf_copy_array() we simply RESOLVE() and then check if obj < LIMIT || obj != ARRAY. | 11:59.16 |
| why coulnd't we do the same in pdf_array_put()? | 11:59.24 |
tor8 | sebras: there's a commit on tor/master (not merged yet) which adds warnings | 12:00.30 |
| if you review the first two commits on tor/master, we can push those and you can work from there | 12:01.10 |
sebras | tor8: oki. give me a minute. | 12:03.34 |
| tor8: so in the first patch override is meant to indicate whether the value returned in advance is valid or not? | 12:09.24 |
| tor8: and therefore whether it is safe to change its sign..? | 12:09.36 |
| tor8: if that is the case shouldn't we bail if override == 1? | 12:10.01 |
| (I have no idea if these advance, uofs and vofs thingies are optional) | 12:10.18 |
tor8 | sebras: they are optional. if override is set, they are provided otherwise they should fall back to their default (i.e. initial, calculated) values | 12:12.11 |
| and if provided, we need to flip the sign in RTL mode | 12:12.30 |
| if defaulted, the sign is already correct | 12:12.43 |
sebras | tor8: ah! | 12:12.49 |
| tor8: but in the case of uofs/vofs no sign needs to be flippe. | 12:13.24 |
| tor8: alright, in that case that patch seems plausible. | 12:13.41 |
| tor8: in the 2nd patch you can warn in pdf_array_delete() | 12:14.43 |
| tor8: you added a fz_warn() further down in the same function. | 12:15.00 |
tor8 | ah! rats. I have an uncommitted diff sitting in my working repo. | 12:15.48 |
| which adds one more warning to pdf_array_delete :) | 12:16.04 |
sebras | tor8: cool, push! I want them! :) | 12:16.18 |
tor8 | sebras: pushed | 12:16.40 |
sebras | tor8: finally! | 12:16.45 |
| tor8: do you want to combine the interval warnings in pdf_array_put() to look similar to those in pdf_array_delete()? | 12:18.35 |
tor8 | sebras: I want to try to make all the warnings exceptions | 12:19.12 |
sebras | tor8: also, how does the trailing warning in pdf_array_put() differ from the one 10 lines above? | 12:19.18 |
tor8 | sebras: gah. that one should have an 'else' | 12:19.54 |
| don't push this commit... rework it to do exceptions and fix my back-from-vacation bugs :) | 12:20.12 |
sebras | tor8: ok, so I take over this commit. can do. :) | 12:20.43 |
tor8 | sebras: if (obj >= PDF_OBJ__LIMIT) is the check for magic pointer values we use for PDF names | 12:20.51 |
sebras | tor8: yes, but we could a test similar to the one in pdf_array_len() everywhere. | 12:22.44 |
| tor8: I don't see why these tests must differ between the functions. (but there might be a subtle thing I'm missing. | 12:23.15 |
| tor8: do you mind re-reviewing my documentation patch? | 13:49.47 |
RobinWattsLenovo | Hi all. | 16:44.18 |
sebras | RobinWattsLenovo: hi! | 16:44.25 |
| RobinWattsLenovo: you found internet! :) | 16:44.30 |
RobinWattsLenovo | yeah, seems very paatchy indeed. | 16:46.05 |
| Can't get mail or facebook. | 16:46.16 |
sebras | RobinWattsLenovo: I'm travelling to shanghai on thursday, I'm not sure they have anything besides qq.cn and baidu.cn..? | 16:47.07 |
RobinWattsLenovo | saw many elephants today. Plus lions, warthogs, various birds, buffalo, waterbeest etc. | 16:47.56 |
| I have a photo of a lion doing a tiger.eps impression :) | 16:48.40 |
sebras | RobinWattsLenovo: are there GNUs there too? | 16:48.41 |
RobinWattsLenovo | :) | 16:50.51 |
| I've got to let helen use the computer now :) | 16:50.51 |
sebras | RobinWattsLenovo: nice to hear that you guys are having a good time! take care and take many photos! :) | 16:51.18 |
| tor8 (for the logs): there are two patches over at sebras/master that 1) turn the warnings for dicts/arrays into exceptions and 2) add stricter checks for these functions. | 20:10.32 |
| tor8 (for the logs): no diffs from trunk. it's 4:30am here. I'm going to school tomorrow, maybe I should sleep. | 20:30.26 |
| Forward 1 day (to 2016/08/23)>>> | |