| <<<Back 1 day (to 2017/06/29) | 20170630 |
Robin_Watts | tor8: First 3 commits on robin/master should be good to go. | 13:30.18 |
| (you may have looked at some of them before) | 13:30.27 |
tor8 | Robin_Watts: those 3 LGTM | 13:32.34 |
Robin_Watts | tor8: Thanks. | 13:33.38 |
mvrhel_laptop | Morning Robin_Watts | 14:49.52 |
| Quick question for you | 14:50.25 |
| with your flipping of cmyk flavor, is pkmraw coming out inverted? | 14:50.50 |
Robin_Watts | mvrhel_laptop: Urm... the halftoning probably needs fixing, yes. | 14:52.52 |
mvrhel_laptop | ah ok | 14:52.57 |
| can you point me to that? Regression testing was a little difficult to judge. Thanks | 14:53.48 |
Robin_Watts | halftone.c | 14:54.37 |
mvrhel_laptop | ok thank you | 14:54.43 |
Robin_Watts | see the note at line 343. | 14:54.52 |
| it should be an easy fix. | 14:54.57 |
| But if you do take it on, please put a FIXME in to remind me to fix the ARM code :) | 14:55.13 |
mvrhel_laptop | Robin_Watts: if it is something you can do in quickly I will let you do it :) | 14:56.19 |
| I suspect there is < that changes to a > | 14:57.14 |
| someplace... | 14:57.17 |
Robin_Watts | mvrhel_laptop: It's exactly that | 14:57.25 |
| Well, >= to < | 14:57.35 |
| in do_threshold_4 | 14:57.49 |
mvrhel_laptop | :) | 14:57.55 |
| ok | 14:57.57 |
| ok I see | 14:58.08 |
| I can do that | 14:58.13 |
Robin_Watts | in the ARM code block all the orrle's change to orrgt :) | 14:58.15 |
mvrhel_laptop | ah ok | 14:58.31 |
Robin_Watts | I'm trying to fix the reference counting problems with colorspaces :( | 14:59.36 |
mvrhel_laptop | sorry. At the time, I did not see any issues with that | 15:00.00 |
Robin_Watts | no, they were all masked before. | 15:00.10 |
mvrhel_laptop | ah ok | 15:00.15 |
Robin_Watts | by removing the is_static stuff, we get crashes, and it exposes a lot of leaks. | 15:00.28 |
| I'll get there. | 15:00.35 |
mvrhel_laptop | ok. let me know if I can help | 15:00.44 |
baskerv | I can't compile the current master branch using lcms2 2.8: https://gist.github.com/baskerville/30cc6b5725daed176a960587ab572e56 | 18:27.16 |
avih | tor8: your modified code has errors on some cases. for instance (0.5).toString(36) is "0". also, it has some trailing 0 after the fraction part, e.g. with (-1.1).toString(2) . also, is there a reason you took the algorithm but not the patch (even if you fix it up a bit)? i did come up with the algo on my own, not that i care too much, but i think thr better thing to do would be to ask me to fix stuff and then take my patch... wouldn't it be? | 18:30.25 |
Robin_Watts | baskerv: No. | 18:30.49 |
| You can't. | 18:30.52 |
| You need to use *our* version of lcms2. | 18:31.00 |
| git submodule update --init and that'll pull it in. | 18:31.20 |
avih | tor8: also, it uses less resolution than it could. for instance (359999999999999.9).toString(36) or (359999999999999.9).toString(8) could have a useful (and accurate) fractional part, but your code makes it an int. TBH though, i'm not quite sure why my limit works better exactly, but it definitely does on every edge case i tested | 18:47.31 |
| OTOH, i think that regardless if choosing 53 or 54, you need to test against the limit after multiplying it by radix to get closer to the limit (on either side). if you first test and then multiply, and e.g. there's 1 fraction _bit_ but radix is much bigger than 2, then you take a digit even from a number which is several bits past the limit | 18:50.38 |
mvrhel_laptop | ok ghent RGB ICC blend mode test working now | 20:15.35 |
| as is the CMYK ICC blend mode test | 20:15.53 |
| There was a bug in soft light | 20:16.03 |
| and the non-separation cases needed to handle CMYK | 20:16.19 |
Robin_Watts | mvrhel_laptop: Hi. | 20:20.15 |
| There are 5 commits on robin/master. The first 2 are ready to go in. | 20:21.01 |
| The second one is yours, the first is mine. If you could review it, I can get it in. | 20:21.26 |
mvrhel_laptop | Robin_Watts; ok looking | 21:12.11 |
| Robin_Watts: which ones am I looking at? | 21:14.58 |
| I see 2 from you then one from me | 21:19.56 |
| I don't see any issues with the Invert subtractive color values commit | 21:28.51 |
| and the separations commit looks ok. The java I am so so at reviewing... | 21:46.37 |
Robin_Watts | mvrhel: Oldest reviews are earliest right? So it's the "Remove is_static..." and "Make sure CMYK ICC color spaces are subtractive" | 22:55.00 |
| Specifically it's the "Remove is_static..." that I need reviewed. | 22:55.23 |
| Forward 1 day (to 2017/07/01)>>> | |