> On Tuesday, April 12, 2016 11:59 AM, Roman Haefeli <reduzent@gmail.com> wrote:
On Fri, 2016-04-08 at 14:38 +0000, Jonathan Wilkes via Pd-list wrote:
Hi list, Here are some binaries to test out the alpha 0 release for the GUI port of Pd-l2ork.
Wow.. impressive work you've done and I think I'm not even remotely able to grasp what it meant to achieve this. I very much like its native look (I checked on Ubuntu 14.04). Also the GUI tab in the preferences dialog, very neat. And the canvas zoom is probably going to be _the_ killer feature with all the hi-res displays coming up.
If you have a chance, do check out doc/4.data.structures/pd-l2ork/ds-tutorialsas well. It's still under heavy development, but I think you'll find the API allows you to do 2d visualizations with a lot less work than the vanilla API. Part of that has to do with affine transformations, which includes zooming (well, scaling)-- that's why I mention it here.
Some observations (from Ubuntu 14.04 i386):
* For a true vanilla experience, it'd be nice if the GUI preset
'vanilla' would use bold fonts.
Added to tracker. I have the feeling I didn't do this because the bold style probably changes the line width, which makes it a harder problem than just adding a line of css. But I'll look into it.
- I sometimes have to click on Menus twice for the selected item to show
up. I couldn't figure out a reliable pattern. Affected entries: 'Media->Test Audio and Midi' or 'Edit -> Preferences' This is a bug in nw.js. It's been reported and investigated, and will hopefully get fixed soon.
- I can't load many of my patches in this nw-version of Pd-l2ork. When I
do so, the patch canvas never appears and I get repeating messages 'watchdog: signaling pd..' on stderr. The process 'nw' uses 100% of a core and I have to kill pd-l2ork. Some simple patches work fine and I haven't figured out a pattern of what kind of patches are affected which are not. However, the behaviour of a certain patch is consistent (either it loads always ok or it never does so). UPDATE: Not true. It just takes that much time to load the patch. The one I just loaded took more than 2 minutes to load. You can see it even with the loadmeter on slower machines-- each additional Pd object triggers a layout recalculation, and that can result in jank. But I think I've got a fix for it. Email me a worst-case patch and I'll test.
What version of Pd-vanilla is this based on? Or is not related to
vanilla anymore? I stopped caring about Pd-l2ork when I figured it contains some fixes that shifts position of iemguis around. This renders a lot of my patches unusable because they are not displayed in the GOP area anymore. I'd love if patches would work in both flavors, but right now it seems as a patch author you need to decide one. Try starting up with the "-legacy" flag and tell me if they display properly. This is a transitional flag-- what I want to do is add a flag to t_canvas so that this "just works" for you and "just works" for Pd-l2ork gops going forward. (And just have some simple mass import for Pd-l2ork gops created in the interim.)
In the meantime, let me know any of your gop abstractions that don't display correctly with -legacy and I'll file a bug and investigate.
Is it correct that externals for Pd-l2ork are binary-incompatible with
externals built for Pd-vanilla? I'm just interested to know, maybe this isn't such a huge issue, since Pd-l2ork has so many externals pre-compiled. Two related issues:1. I am porting gui externals manually to nw.js. In these cases binary compatibility is irrelevant.2. AFAICT nobody has had sufficient interest to test whether or not Pd-l2ork and Vanilla are binary compatible. So I think I'll just blithely claim they are 100% binary compatible. If anybody finds otherwise, I'll file a bug. -Jonathan
Roman
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list