| <<<Back 1 day (to 2017/10/29) | 20171030 |
sebras | Robin_Watts: did you give up on the indeterminisms for today? | 14:46.36 |
| tor8: two silly patches on sebras/master | 14:46.44 |
Robin_Watts | sebras: No. I'm hoping to get to them. | 14:46.49 |
| I'm battling the cluster still. | 14:46.56 |
sebras | Robin_Watts: ok, then I'll not bother you. :) | 14:48.01 |
Robin_Watts | sebras: First one lgtm. | 14:51.21 |
| second one... is buf now returned uninitialised? | 14:51.41 |
sebras | Robin_Watts: FT_Get_Glyph_Name() starts by checking that face != NULL and then that buf != NULL and then that size != 0 | 14:53.45 |
| next it sets buf to '\0'. | 14:54.03 |
| so I think it is reasonably safe? | 14:54.08 |
Robin_Watts | Ah, insuffient context in the diff :) | 14:54.17 |
| lgtm then. | 14:54.19 |
sebras | we do similarly in other places. | 14:54.32 |
| we never checked the return value in this one place though, which prompted coverity to complain. | 14:54.59 |
Robin_Watts | sounds ideal. I just couldn't see the *buf=0 call in the git web view. | 14:55.04 |
sebras | np. | 14:55.10 |
| thanks! | 14:55.16 |
Robin_Watts | tor8, sebras, mvrhel_laptop: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=80ef17d3386280e05287d61f94677d4b79103cf4 | 16:32.59 |
| That fixes the interminisms I think. | 16:33.06 |
tor8 | Robin_Watts: doesn't seem unreasonable to me | 16:37.14 |
sebras | looks. | 16:37.23 |
Robin_Watts | yeah, the bmpcmp gives what I'd expect. | 16:38.25 |
sebras | Robin_Watts: that explains it, my src == RGB my dst == CMYK+2 spots. | 16:47.10 |
Robin_Watts | sebras: The case I caught in the debugger, was src = grey, with 2 disabled spots, dst = grey | 16:48.02 |
sebras | disabled == COMPOSITE? | 16:48.19 |
Robin_Watts | yeah. | 16:48.32 |
sebras | Robin_Watts: I have a question. | 16:52.01 |
| Robin_Watts: in push_group_for_separations you set the default colorspace to CMYK and clone the separations. | 16:52.17 |
| the separation are in this case changed COMPOSITE->SPOT... why? | 16:52.37 |
Robin_Watts | In order to do overprint simulation, we need to render spot colors separately. | 16:53.04 |
| if you're mapping a spot down to process colors, you can't then remove it when someone overprints it later :) | 16:53.38 |
sebras | reads that sentence slowly and carefully... | 16:54.31 |
Robin_Watts | Suppose i have a file with CMYK + Orange + Green. | 16:55.30 |
sebras | y | 16:55.36 |
Robin_Watts | I draw an orange rectangle on the backdrop. | 16:55.43 |
| Then I 'overprint' a Green rectangle on top. | 16:55.59 |
sebras | y | 16:56.10 |
Robin_Watts | depending on the OPM and OP values, I may need to remove the CMYK values, but not the Orange values when I plot the green rectangle. | 16:56.44 |
sebras | ok, so COMPOSITE in this case means that Orange in this case would be mapped to some specific levels of CMYK, then later when you print the Green rect you'd remove all the Orange, which was not supposed to happen. | 16:58.50 |
| but if it is SPOT then the Orange rect would be kept separate and you can remove whatever CMYK values are below..? | 16:59.39 |
Robin_Watts | yes. | 17:00.41 |
sebras | ok, thanks. | 17:00.54 |
Robin_Watts | no worries. | 17:01.00 |
sebras | when I debugged this the other day I never noticed that the behaviour setting affect a clone of the separations struct, and so was thoroughly confused. | 17:01.20 |
Robin_Watts | overprinting is the primary reason we push the group. | 17:01.33 |
| (though it is required for different base color spaces too, but that doesn't require the new spots channels) | 17:02.06 |
sebras | Robin_Watts: UMR? | 17:10.16 |
Robin_Watts | Uninitialised Memory Read | 17:10.26 |
sebras | LGTM, and fixed Transparency-DesignGuide.pdf that I tested. | 17:11.23 |
| (I ran it in ASAN) | 17:11.33 |
Robin_Watts | sebras: Thanks. | 17:11.53 |
| Forward 1 day (to 2017/10/31)>>> | |