| | <<<Back 1 day (to 2017/04/17) | 20170418 |
| kens | And I thought we'd been promised we would be able to do the documentation ourselves | 14:43.40 |
| | OOps wrong channel :-( | 14:43.52 |
| Robin_Watts | Ok, so the reported problem with the crashing in the scan converter boils down the way I'm doing bezier subdivision. | 18:59.35 |
| | fixed ax = (sx + c1x)>>1; | 18:59.46 |
| | if sx and c1x are both 0x800000000 then we overflow and get 0. | 19:00.08 |
| | I can fix it by doing: fixed ax = (sx>>1) + (c1x>>1) | 19:00.26 |
| | but that means we are throwing away accuracy in the common cases. | 19:00.41 |
| | I could detect the overflow case and use a separate 'big' code path. Any opinions? | 19:02.03 |
| ray_laptop | mvrhel_laptop: oh, I should chat here since it is a ghostscript question | 19:24.59 |
| mvrhel_laptop | ray_laptop: any channel is fine | 19:25.45 |
| ray_laptop | So I have a file that has an SMask that then pushes a transparency group. The SMask pushes a buffer that doesn't have tags, fine (the device ENCODES_TAGS), but the transparency group buffer uses the tos->has_tags and is DeviceCMYK, but the encode_color is still pdf14_encode_color_tags | 19:27.48 |
| mvrhel_laptop | ray_laptop: "So I have a file that has an SMask that then pushes a transparency group" So are you in a soft mask group drawing and there is a transparency group pushed while you are inside the soft mask? | 19:31.19 |
| ray_laptop | mvrhel_laptop: yes, sorry for the awkward wording. | 19:32.02 |
| mvrhel_laptop | Or has the soft mask been drawn and is in the buffer, ready to be composed upon the next group? | 19:32.12 |
| ray_laptop | no, it's still in the SMask | 19:32.37 |
| mvrhel_laptop | ok. So we are busy drawing away in our soft mask buffer. I see. We really should be changing the encode color stuff in this case | 19:33.43 |
| | We don't want any tag updates while we are drawing in the softmask | 19:34.00 |
| | So any group pushes that occur while we are active in a softmask drawing need to be handled special | 19:34.33 |
| | starting to feel like I have a perfect storm brewing this week. | 19:35.27 |
| | Soft masks, color management with threading in mupdf, kitchen work to resume, hard wood floors getting refinished and wife's birthday all this week | 19:36.48 |
| | ray_laptop: Do you want to drop this on me or do you want to give it a try to fix. I know I am suppose to look at something color group related for ken | 19:37.50 |
| | Maybe I should stop mupdf for a couple days and look at my gs stuff | 19:38.08 |
| ray_laptop | mvrhel_laptop: I just wanted to see if that rang any bells. It looks like it is ignoring the tag even though it's in the color since tha mark_fill_rectangle uses the has_tags from the buf structure (which is 0 in both the SMask buf and the CMYK group buf) | 19:39.20 |
| mvrhel_laptop | oh well that is good. It should ignore the tags shouldnt it? | 19:39.52 |
| | ray_laptop" ^^ | 19:40.03 |
| ray_laptop | mvrhel_laptop: I just wanted to see if it rang any bells, but let me keep grinding on it -- I'm getting bad colors from somewhere with this file. | 19:40.33 |
| mvrhel_laptop | ray_laptop: let me know if you need me to help out | 19:44.40 |
| ray_laptop | OK, | 19:45.00 |
| | Forward 1 day (to 2017/04/19)>>> | |
Log of #ghostscript at irc.freenode.net.