 | PGF/TikZ | 56 Members |
| https://github.com/pgf-tikz | 11 Servers |
| PGF/TikZ | 56 Members |
| https://github.com/pgf-tikz | 11 Servers |
| Sender | Message | Time |
|---|---|---|
| 8 Oct 2021 | ||
 hmenke | * | 14:43:37 |
 samcarter | In reply to @hmenke:matrix.orgThanks for taking a look! | 14:53:16 |
| samcarter | the cmyk shading problem seems to first occur in TL2019, with TL2018 from overleaf it seems to work fine | 14:57:24 |
| samcarter | ... seems related to the new shadings colour space thingy from 65da52fd74426af68a42dcf5bffacf0aec209a1a | 15:07:01 |
| samcarter | ... guess I'll create an issue about this | 15:10:50 |
| hmenke | Well, before that all shadings were just RGB, which is a different problem. | 15:16:50 |
| samcarter | The easiest solution would just to abolish cyan - for the other colours one does not really notice a difference between the models :) | 15:21:09 |
 Ulrike Fischer | samcarter: what is the problem with cmyk? It looks ok to me. | 16:10:16 |
| 11 Oct 2021 | ||
| hmenke | muzimuzhi: What do you think about requiring Conventional Commits for PGF? | 10:12:30 |
| hmenke | josephwright: I've experimented with flattening the doc tree: https://github.com/hmenke/pgf/commit/4e30e38136b9e6acda47975525596c004fade355 | 17:59:53 |
| hmenke | It works okay but there are some problems with usability. | 18:00:11 |
| hmenke | The aux files are not reusable between the different engines. | 18:00:37 |
| hmenke | So after compiling for, e.g. LuaTeX all aux files have to be deleted before being able to compile with. e.g. tex4ht. | 18:01:14 |
| hmenke | There error messages that this causes are also not exactly useful. | 18:01:36 |
| hmenke | * The error messages that this causes are also not exactly useful. | 19:18:56 |
 josephwright | In reply to @hmenke:matrix.orgWell yes, that would be expected: l3build expects a single engine to be used for docs, or for things to be cleaned up if you make major changes | 21:36:26 |
| josephwright | In reply to @hmenke:matrix.orgIt's aimed at automation: stuff that's out-and-out broken tends to require manual work first | 21:36:51 |
| josephwright | In reply to @hmenke:matrix.orgI've got some other stuff to look at for a few days: I'll try to get back to this after that (I want to read Marcel Krüger's suggestion) | 21:37:23 |
| 12 Oct 2021 | ||
 muzimuzhi | In reply to @hmenke:matrix.orgThat's good and will help train contributors, including me, into real developers. Though I think insufficient contributions is another problem that PGF encounters. | 04:47:02 |
| muzimuzhi | In reply to @hmenke:matrix.org Maybe https://ctan.org/pkg/unravel can help, but some further development based on it is needed. I've been using The cons are, an extra level of abstraction, inactive maintainers (@PhelypeOleinik may already get some familiarity about the code base, according to this issue comment), and the difficulties of debugging | 04:59:00 |
| muzimuzhi | In reply to @hmenke:matrix.org* That's good and will help train contributors, including me, into real (or better) developers. Though I think insufficient contributions is another problem that PGF encounters. | 09:45:19 |
| josephwright | In reply to @muzzi:matrix.orgBoth Phelype Oleinik and I know something about the code: I guess we should fix the issues :) | 09:45:58 |
| hmenke | muzimuzhi: unravel is a very impressive effort but unfortunately it is very slow and only helps with expansion. For instances where I have to find a spurious space I cannot get around \tracingall. It would be nice if there was some engine feature to filter the logs. | 10:42:04 |
| muzimuzhi | Yes, it's not a silver bullet 😮💨 | 10:43:44 |
| hmenke | Or something to exempt certain macros from tracing altogether and only resume tracing when the expansion of that has finished (looking at you \pgfmathparse). | 10:44:06 |
| hmenke | Although, this might actually do the trick for | 10:46:17 |
| hmenke | This of course does not work for stuff that has to be expandable like \pgfutil@ifnextchar or so. Most of the time I don't care how it figures out which branch to take but only which branch is taken. | 10:50:01 |
| muzimuzhi | In reply to @hmenke:matrix.orgDeciding whether a macro has "finished" is non-trivial i guess. The info of where a token comes from is required. Expansion of a macro? \aftergroup? Read from input file? ... | 10:52:29 |
| hmenke | I don't think it's non-trivial. When TeX sees \pgfutil@ifnextchar it reads #1#2#3 from the input and starts expansion. As soon as TeX reads the next token from the input it means that \pgfutil@ifnextchar has finished. | 10:54:03 |
| hmenke | That's something that should be quite easy to introspect on the engine level. | 10:54:54 |