ID: | 5091 |
From: | rn...@igor2.repo.hu |
Date: | Wed, 18 Aug 2021 04:06:16 +0200 (CEST) |
Subject: | Re: [pcb-rnd] long term plans: HIDs, sdl, gtk4 (fundrising or |
in-reply-to: | 5089 from Alain Vigne <al...@gmail.com> |
replies: | 5092 from Bdale Garbee <bd...@gag.com> |
On Tue, 17 Aug 2021, Alain Vigne wrote: >No, GL using hardware acceleration is very usable... as is Cairo when >hardware acceleration is possible, and enabled ! And then we don't need cairo. We have direct opengl render code. Cairo is nothing else just an unneded, excess software layer between two other layers of code. Such things typically don't speed things up but slow things down. <snip> >Now, I think GTK4 (note it is not GTK+-4.0 ;) Does not matter at all. I am not obsessed with gtk, I hate it. For me it doesn't matter what random differences they do in their totally broken release cycles/versioning. I have zero trust in gtk dev team ability to maintain _any_ sort of API stability or reasonable releases. At the end, if the community decides to support the effort, there's only one thing that matters: to use the latest release that seems to hit distros to maximize the number of years on gtk we buy for our money/time. I will look at this one thing if we ever get there, and use the matching version, whether they will call it gtk5, gtk16 or "gtk 3.9999.9.99 that is not gtk+-3.0" >is still using Cairo for all >rendering, but using Wayland, hardware acceleration is active. So, probably >I would have better results now, who knows ? It probably uses cairo for rendering the widgets. Rendering speed is not critical for buttons, it's critical for rendering the board. You wouldn't have better results now: for the critical path, board rendering, cairo is still: - unacceptably slow in sw rendering - just an unwanted extra layer, a slowing factor in the hw rendering path that adds redundant code (compared to simply using the already existing direct opengl code) > >There is a clear choice to be made for the main PCB display: Use OpenGL API >or Cairo one. That is the difference we already have between hid_gtk2_gdk >and hid_gtk2_gl. >Igor2 states : use OpenGL. I am fine with this. No pb. >But still all the GTK4 widgets changes + new mechanisms are troublesome to >get a GTK4 HID. But why do you care? What do you have to do with all this? You stated you won't donate to the cause. I started this thread for one reason, and made it explicit right from tbe beginning: to see if the community funds the project or not. It never was a generic "vote if you like gtk4" thread. It never was a "help me figure gtk4 details". You don't donate and you won't contribute to the coding, so for you this would be just yet another random plugin that just happens independently from you. If it happens at all... So let's close the technical detail part of this discussion, as especially in this stage it is largely irrelevant. We should focus on what's more important in this subproject's life: whether it starts/happens at all, or whether I do the SDL HID in the same time slot this winter instead. At the moment gtk4 is not winning this time slot yet. Majenko, Bdale, anyone else who think they prefer gtk4 over sdl? Best regards, Igor2
Reply subtree:
5091 Re: [pcb-rnd] long term plans: HIDs, sdl, gtk4 (fundrising or from rn...@igor2.repo.hu
5092 Re: [pcb-rnd] long term plans: HIDs, sdl, gtk4 (fundrising or from Bdale Garbee <bd...@gag.com>
5093 [pcb-rnd] gtk4 conclusion (was: Re: long term plans: HIDs, sdl, gtk4 (fundrising or contribution)) from rn...@igor2.repo.hu
5094 Re: [pcb-rnd] long term plans: HIDs, sdl, gtk4 (fundrising or from John Griessen <jo...@cibolo.com>
5095 Re: [pcb-rnd] long term plans: HIDs, sdl, gtk4 (fundrising or contribution) from Evan Foss <ev...@gmail.com>