| <<<Back 1 day (to 2016/08/06) | 20160807 |
fredross-perry | sebras, robin - one final question. If I call jni_throw, as I'm doing with allocation failures, do I also need to return NULL? | 00:56.28 |
| after a little reading, the answer would seem to be Yes. | 01:02.47 |
sebras | fredross-perry: yes, return is necessary. I forgot to check that last time. :/ | 04:46.10 |
| fredross-perry: you call fz_stext_char_bbox() twice which appears to be unnecessary. | 04:46.34 |
fredross-perry | the returns are there. Just confirming that they are necessary. Thanks. | 04:46.39 |
| so i do. nice catch thanks. | 04:47.52 |
sebras | fredross-perry: if we throw at any given point, do we also want to clear StructuredText.blocks? So as not to leave it halfway initialized? | 04:54.30 |
| fredross-perry:corner case, I know. But better to think about now rather than later when debugging a user report. :) | 04:55.06 |
| fredross-perry: I'm not sure I handle this correctly myself everywhere. :/ | 04:55.43 |
fredross-perry | yes I think we would. Do we like/hate the idea of using goto? I'm thinking about readability. We do it in SOL. | 04:56.55 |
sebras | fredross-perry: I think gotos are used in several places. | 04:57.47 |
fredross-perry | I'll take a look at that tomorrow night. | 04:58.31 |
| thanks | 04:58.36 |
sebras | I was unfortunately correct. I found two places whre arrays are constructed and stored but the elements in these arrays may only be halfway constructed. | 05:10.03 |
| Robin_Watts: so... there is a nice patch to review over at sebras/master :) | 05:10.15 |
| Robin_Watts: actually, you get one more patch which fixes fz_vsnprintf()'s handling of %zu on 32-bit platform (hello!) | 05:14.49 |
| fredross-perry: oh, and a trivial thing: keep all the fid_-declarations together. I think they are also meant to be sorted. | 05:16.23 |
| Forward 1 day (to 2016/08/08)>>> | |