fix negative thickness of polylines in 4xxx.lib #1059
Conversation
|
|
|
@poeschlr |
|
@evanshultz @poeschlr I'm open for feedback. The violations that travis reports were not introduced with this commit. |
antoniovazquezblanco
added
Enhancement
|
This has already caused 2 duplicate bug reports being opened over at launchpad. |
|
I am open to fixing this on the symbol side. But it should also be fixed in opengl. However the value of -1000 was choosen to result in no outline shown. 10 mil is therefore too much. I would go with the lowest value that is not replaced with the default. So 1 mil. |
Changed to width 1
|
Ok, changed to 1 mil, but I doubt that's what users want... |
|
https://www.mail-archive.com/kicad-developers@lists.launchpad.net/msg32090.html Those messages discuss doing exactly this, using 1mil line thickness. But the message after it says that Jeff backed out a change. I do not know what version of KiCad you're using and if the rendering you're showing is how it will be in 5.1. What build are you using? Did you change the symbol in any way to produce the above screenshot? I see a 10mil horizontal line on top and bottom of 4001 to provide a consistent symbol outline thickness. I also see the background filled. Neither of this appear above, but I don't know if that is solely the result of the new GAL. Last, note that we should be careful because this library will be used for all KiCad users during the 5.0 cycle before 5.1 is released. I did a quick check and I don't see any issues with 1mil lines in 5.0, but note that most of them are covered up by a 10mil line. |
|
I'm using: Build settings: I had background set to white, so that's why.. Changing it to some yellowish color gives me: |
|
PS: Maybe the confusion is that I was using latest on master, versus 5.0.0 or 5.0.1 So one could argue, that I'm ahead of the released versions and that's to be fixed later, but it causes bugs being reported among nightly users... I dunno |
|
Yes, I understood that and it's why I'm not in a panic. It is not clear from the above discussion how the GAL renderer will perform. Using 1mil was suggested by JP but I'm uncomfortable merging this since so many things are up in the air now and we're quite far from even a RC1. And since this doesn't affect 5.0 and is not totally fixing 5.1 I don't see a great impact. If @poeschlr wants to merge or ask me to merge that's fine, since it is some incremental improvement for master, but there's still figuring out to do. |
|
I agree with @evanshultz that we might want to hold of at least till the feature freeze of 5.1. |
|
Should we use the |
|
While this was definitely an issue, I believe it was one of the teething pains of moving Eeschema to the new rendering engines. The discussions mentioned earlier to the devs appears to have resulting in a fix over the last few months. Here is a recent nightly using the Modern Toolset rendering: I think we can close this PR. (But we should definitely make an update of the logic symbols a major target of the v6 libs IMO.) Application: kicad Build settings: |
myfreescalewebpage
added
Abandoned
tl;dr, negative line thickness creates awful filled polygons.
See https://bugs.launchpad.net/kicad/+bug/1797283 for the genesis of this patch.