Blog: On Patches

Here you can discuss ReactOS related topics.

Moderator: Moderator Team

Post Reply
Z98
Release Engineer
Posts: 3379
Joined: Tue May 02, 2006 8:16 pm
Contact:

Blog: On Patches

Post by Z98 » Mon Dec 15, 2014 1:11 am


PurpleGurl
Posts: 1774
Joined: Fri Aug 07, 2009 5:11 am
Location: USA

Re: Blog: On Patches

Post by PurpleGurl » Mon Dec 15, 2014 7:21 am

Yes, I noticed in the logs that Z98 was getting involved. I thought that was odd since I hadn't seen much if any activity in the past, and I remembered him saying in the past that he had reasons for not being a developer. So it was a refreshing change to see that. Much of that isn't big stuff, but someone has to do it, and all the little stuff does add up and contribute to the whole. The thing is that so many see it as little and push it aside, and then there is a backlog of things to do, and it makes it even more ominous.

I noticed the part about GDI and GDI+. It would be nice if some of that code could be improved beyond Microsoft in terms of performance. I read that GDI+ in Windows is not hardware accelerated and does more text stuff and handles larger resolutions, and thus is 1/6 the speed of GDI in a number of cases. Then Vista and 7 uses DWM/WDDM, and hardware acceleration still isn't used. So it would be nice if we could do it using acceleration where it would help. I know that acceleration does not always help, since just like DMA on older machines, acceleration is often used to reduce CPU usage, not speed up things, and may sometimes slow things down. In the case of video codecs, if 2-way operation is necessary, it can slow things down unless the card and drivers were made to be accelerated in both directions. Otherwise stuff get copied into system memory and worked on by the CPU rather than the GPU, and would in that case be faster using the CPU alone. I know I am not saying all this right. But like I said, if using acceleration could be done effectively when able, we could have an advantage over Microsoft.

Kargaroc586
Posts: 8
Joined: Wed Apr 30, 2014 7:42 pm

Re: Blog: On Patches

Post by Kargaroc586 » Mon Dec 15, 2014 8:57 am

PurpleGurl wrote:I noticed the part about GDI and GDI+. It would be nice if some of that code could be improved beyond Microsoft in terms of performance. I read that GDI+ in Windows is not hardware accelerated and does more text stuff and handles larger resolutions, and thus is 1/6 the speed of GDI in a number of cases. Then Vista and 7 uses DWM/WDDM, and hardware acceleration still isn't used. So it would be nice if we could do it using acceleration where it would help. I know that acceleration does not always help, since just like DMA on older machines, acceleration is often used to reduce CPU usage, not speed up things, and may sometimes slow things down. In the case of video codecs, if 2-way operation is necessary, it can slow things down unless the card and drivers were made to be accelerated in both directions. Otherwise stuff get copied into system memory and worked on by the CPU rather than the GPU, and would in that case be faster using the CPU alone. I know I am not saying all this right. But like I said, if using acceleration could be done effectively when able, we could have an advantage over Microsoft.
Especially nowadays when you usually have no non-3d hardware acceleration whatsoever in >=Win8
Though that'll take a loooooong time before its possible in ROS

Tonix
Posts: 89
Joined: Tue May 22, 2007 11:33 am

Re: Blog: On Patches

Post by Tonix » Mon Dec 15, 2014 12:21 pm

Thank you for review the patches submitted,
i look at jira at lest once a week, to provide some debug log, or to report some issue,
and i ask myself many times why most patches were ignored.

I would like to point to another problem, similar to this, really frustrating.

Most of the older bug are still assigned to the user "Bug Zilla" and no one seem to care about them.

I think review them is needed and should be piority over developing new function.

Reactos needs more stability.

To reach stability you need closing old bug, if possible it's.

Have 1100 issues over 1500 (over 70%) assigned to "no one" is a non sense.

PurpleGurl
Posts: 1774
Joined: Fri Aug 07, 2009 5:11 am
Location: USA

Re: Blog: On Patches

Post by PurpleGurl » Mon Dec 15, 2014 3:47 pm

Kargaroc586 wrote:Especially nowadays when you usually have no non-3d hardware acceleration whatsoever in >=Win8
Though that'll take a loooooong time before its possible in ROS
I am afraid so. As for why Microsoft doesn't do it, I can think of a handful of ideas. I know they've improved the logic, and thus in some cases, it is faster even without acceleration. Older Windows versions used to redraw more. And there are cases where hardware acceleration can introduce incompatibilities or slow graphics down. And it may be a bit difficult to code for, so adding all that together, they might not use it. It not being used might be what gives Windows 8 a perception of bloat to some people, since the CPU would be used more for graphics.

For the other things discussed, I am glad to see that a lot of old patches are finally being reviewed where applicable. That is something that needs better care and handling. I remember the memory manager regression, and there was a patch sitting around for 3-6 months, and then I think about the time the devs started to get an idea of the cause, the patch resurfaced, and it was all kind of embarrassing. So it seems like there needs to be better coordination or something.
Tonix wrote:Reactos needs more stability.

To reach stability you need closing old bug, if possible it's.

Have 1100 issues over 1500 (over 70%) assigned to "no one" is a non sense.
I agree. It needs better stability and compatibility. One thing that would help would be to work on everything where a blue screen, memory/resource leak, or application crash has been reported.

Webunny
Posts: 1201
Joined: Sat Apr 28, 2012 1:30 pm

Re: Blog: On Patches

Post by Webunny » Mon Dec 15, 2014 11:58 pm

"Another big patch is actually by Hanke, and it's to switch the wallpaper rendering from using GDI to GDI+. Why does this matter? Well, for one it would allow for using images in formats besides BMP as wallpapers."

:o

What the..?! This issue has been mentioned on the forum several times by now, not even that long ago! :shock: And there was a patch for it?

Well, I guess better late than never, and it would help a lot in that department at least. For each BMP you could place 2-3 jpg or npg's in there, and still be lower on MB.

I'm also glad to see this sort of 'JIRA scrutiny'. Good work, z98. It was one of the major weak points, as I've pointed out before, and I'm glad it's recognised. In fact, I suggested in the past, if the (next) IGG goes well, or there is enough money and Giga has finished his work, to hire him or another developer to specifically do exactly as you now are (temporarily?) doing: constant checking up on JIRA bugs, comment on them or ask for further testing or whatever, look at the patches, and communicate with or even badger the respective devs if it's something out of his domain.

There is no doubt the follow-up of JIRA-bugs and patches alike is pretty low, and yet, it's always the first thing that is asked of testers. I've noted myself that, if I hadn't mentioned some of my bugs on irc, they would never have been looked at, let alone be resolved. As said, it's a weak point of the development process in ROS, and yet it's rather an important one, since it directly influences the coding and the workability of the program.

I'm well aware the devs prefer to work on their own part and/or move on to new things within their domain, but bug-squashing and patch-reviewing are quite important too. Putting a dedicated force (well, person) on that would be worthwhile.

fulea.stefan
Posts: 121
Joined: Sun Aug 14, 2011 3:38 pm
Location: București (Bucharest), România

Re: Blog: On Patches

Post by fulea.stefan » Fri Dec 19, 2014 2:21 pm

Webunny wrote:I'm well aware the devs prefer to work on their own part and/or move on to new things within their domain, but bug-squashing and patch-reviewing are quite important too. Putting a dedicated force (well, person) on that would be worthwhile.
The patch-review in itself can be a process of evaluating the quality (and dedication) of some contributors (like Ricardo Hanke or Lee Schroeder) in order to elevate some of them to the ranks of ReactOS developers, if they're interested.

dsp8195
Posts: 86
Joined: Fri Feb 07, 2014 5:35 am

Re: Blog: On Patches

Post by dsp8195 » Mon Dec 22, 2014 3:11 pm

Patches definitely need more love.

florian
Posts: 453
Joined: Tue Nov 01, 2005 2:19 am
Location: Germany

Re: Blog: On Patches

Post by florian » Wed Dec 24, 2014 12:00 am

Make code, not love ;)

gonzoMD
Posts: 1019
Joined: Fri Oct 20, 2006 7:49 am
Location: Germany
Contact:

Re: Blog: On Patches

Post by gonzoMD » Wed Dec 24, 2014 11:51 am

I also saw the great effort, Ricardo did to fix our usermode applications. So in the past I emailed him if he is willed to help out to implement the new application manager. After one day of decision he said yes and then I wanted him to talk to victor or amine but when he was there none of them was available. I don't know whether any further efforts were done to contact him. He is a motivated developer who really wants to help out in things where the core devs don't have the time for. I think another try should be done.

Post Reply

Who is online

Users browsing this forum: DotBot [Crawler] and 1 guest