| | <<<Back 1 day (to 2018/12/13) | 20181214 |
| sebras | Robin_Watts: may I ask you for a review of sebras/master? | 14:02.30 |
| Robin_Watts | sebras: Sure. | 15:05.23 |
| | hmm. | 15:06.19 |
| | I thought that if we did an fz_convert_color to grayscale, then we should only need to give a single space in the array. | 15:06.47 |
| | Yeah, having looked at the code, I still think that. | 15:08.17 |
| | Presumably you have examples where that isn't the case? | 15:08.31 |
| | sebras: Second one looks fine. | 15:14.02 |
| | I hadn't considered that an icc colorspace would need an icc alternate. | 15:16.00 |
| | let me rephrase. | 15:16.10 |
| | I hadn't considered that an icc colorspace could every have an icc alternate, but I guess it's possible and permitted. | 15:16.27 |
| sebras | Robin_Watts: the issue with convert color happened when you end up in icc_conv_color() where link == NULL is interpreted as being gray->cmyk | 15:16.31 |
| Robin_Watts | Third one looks plausible. | 15:16.43 |
| sebras | and so if you only have a single float dstv then you overwrite your stack. | 15:16.46 |
| | I had a file at the time. but where did it go? | 15:17.00 |
| Robin_Watts | sebras: Then that sounds like a bug in that we should never end up there. | 15:17.01 |
| | sebras: I was meaning to say that mentioning files in the commit messages rather than using pithy 1 liners would be good. | 15:17.38 |
| | I know I'm not perfect in that regard. | 15:18.08 |
| | (but in every other way, obviously :) ) | 15:18.21 |
| sebras | I think all of this originated with me attempting to use pwg output and then displaying the PDF generated with cups. | 15:19.12 |
| | Robin_Watts: boids3-page1.pdf is what triggered me to do the third commit. | 15:32.51 |
| | in /home/sebras on | 15:33.07 |
| | casper | 15:33.08 |
| Robin_Watts | It's the first commit that's interesting. | 15:33.13 |
| sebras | I don't remember how I triggered it. :( | 15:33.48 |
| Robin_Watts | Then I wouldn't commit it. | 15:35.01 |
| sebras | Robin_Watts: I never commit anything that doesn't pass the tor/robin-filter. | 15:37.49 |
| Robin_Watts | likewise me and the tor/sebras filter :) | 15:38.15 |
| sebras | you mean the tor-filter and the sebras-sieve? ;) | 15:39.29 |
| | is there a LGTM on the other two? | 15:40.01 |
| Robin_Watts | yes. | 15:40.32 |
| sebras | ta | 15:41.15 |
| | Robin_Watts: found it! :) | 15:56.43 |
| | so clustering sebras/master~1 caused segfaults in GWG1610_Softmasks_Text_part1_X4.pdf | 15:57.01 |
| | this is however fixed with sebras/master~0 | 15:57.44 |
| | _that's_ why I thought it was related to the icc alternate cspace thingy. | 15:58.07 |
| Robin_Watts | right, so I'd like to understand how we ever end up in that code. | 16:00.41 |
| | If we're going to a device_gray space, then we should only ever need 1 float. | 16:01.02 |
| | if we're going to rgb, we should only ever need 3. | 16:01.12 |
| | Forward 1 day (to 2018/12/15)>>> | |
Log of #mupdf at irc.freenode.net.