Revision 53952

[PurpleGurl]

I just burned a live CD and tried it on real hardware. I was able to start it all the way with no problems, and without changing any CMOS settings. I noticed there was no mouse cursor in stage 2. That is a difference from erratic mouse behavior in stage 2, though either way is trivial. In stage 3, the mouse cursor was there and acting almost completely normal. I say almost because it sometimes flickers or disappears when you move it, but that seems inconsequential.

Then I went to mess around in the device manager. While what showed up in there appeared reasonable, I noticed a lot of things has yet to be written in there. For instance, none of the properties tabs has any adjustable settings, even where they are expected under Windows. Over under the virtual memory and paging settings, no drives showed up.

Then I was curious to try programs, but I was unable to due to problems. First, there were no desktop icons at all. My Computer was nowhere to be found. So I figured I'd use the drive browse on the start menu. Fine, except the delay speed on previewing the contents was too short, making it impossible to move the mouse pointer to the correct drive without some other drive with lots of contents opening and making it impossible to move the focus to another drive. And I would click elsewhere and wait for that to leave, and unsuccessfully attempt again to get the mouse to the correct drive. So I decided to try a 3rd route to where I wanted to go, and that was through the Run dialog. I tried a number of different drives, only to get the explorer pulling up with some exception. I think it was something about one of the header files and it gave a line, but I didn't have a pen and paper handy. Not being able to do anything I wanted to try, I shut down the machine.

[vicmarcal]

odd
maybe you are suffering the same issues than Jimtabor has...and it is a srious regression

[PurpleGurl]

odd
maybe you are suffering the same issues than Jimtabor has...and it is a srious regression

Do you have any links to that? And should I go looking for the regression?

[vicmarcal]

Info of the bug:
http://reactos.org/bugzilla/show_bug.cgi?id=6271

Maybe it should be nice to create a list of all the devices your computer has in.That can help to find the "guilty" component if any. Finding the guilty revision, if any, would be quite useful.Try first if 0.3.13 works perfectly in your RH.

[PurpleGurl]

Info of the bug:
http://reactos.org/bugzilla/show_bug.cgi?id=6271

Maybe it should be nice to create a list of all the devices your computer has in.That can help to find the "guilty" component if any. Finding the guilty revision, if any, would be quite useful.Try first if 0.3.13 works perfectly in your RH.

Mine ran 0.3.1.3 fine. I was one of the ones who helped give the leads to resolve the partition problem where it only mounted every 4th drive (ie., C:, G:, K:, etc), bug 2564. The last trunks I tried were maybe 52678, 52731 (mouse regression, so couldn't test the partition fix), and 52733 or 52749 and discussed in this thread: http://www.reactos.org/forum/viewtopic.php?f=14&t=9419

So, I've been able to run it on real hardware without missing icons or the explorer exception for some time.

I just went to the link, but that was about the mouse, not the explorer nor missing icons. So the mouse didn't let me control it during 2nd stage. I don't see the big deal there, since it works fine once booting is complete. But what I noticed was just a fluke, as I could control it while drivers were still loading in my test run since then. In addition, the bug you gave seems to be unrelated due to the scope of revisions given. Back then, it worked erratic during 2nd stage, but normal during 3rd. But the first time I tried 53952, I assumed the inability to move the mouse was by design. Windows sometimes won't let you have it that early either.

Still, even now, as back then, I must have legacy support disabled for the mouse to work, even in stage 3. I am glad you guys removed the broken 2.x support. That makes it possible to test Reactos on real hardware and then boot into Windows without editing the CMOS, and Windows doesn't care either way on the legacy USB setting.

I am much more concerned with the missing icons and the explorer exception. I'd like to find out where the explorer (and possibly icons too) broke. That seems like a blocker bug to me. If you cannot open drives or folders, that is quite serious.

If you want to know about my hardware:
2.00 gigahertz AMD Athlon 64 X2 Dual Core
128 kilobyte primary memory cache
512 kilobyte secondary memory cache
64-bit ready
Multi-core (2 total)
Not hyper-threaded

Board: ASUSTeK Computer INC. M2N-E SLI 1.XX
Bus Clock: 200 megahertz
BIOS: Phoenix Technologies, LTD ASUS M2N-E SLI ACPI BIOS Revision 1204 09/18/2008

NVIDIA MIRROR 298.09G [Hard drive] (320.07 GB) -- drive 0
319.95 Gigabytes Usable Hard Drive Capacity
41.73 Gigabytes Hard Drive Free Space

SONY DVD RW DRU-720A [Optical drive]
TigerJet HardDisk USB Device -- drive 1 2048 Megabytes Usable Installed Memory (Magic Jack phone)
3.5" format removeable media [Floppy drive]

Slot 'DIMM_A1' has 1024 MB
Slot 'DIMM_B1' has 1024 MB
Slot 'DIMM_A2' is Empty
Slot 'DIMM_B2' is Empty

[frmariam]

I've been getting the same explorer issues with livecd builds from over a week or more... Though I thought it was related with me running it in MobaLiveCD (a small self contained qemu vm app... I like for it's size portability and ease of use even though it lacks configuration).

I found that toggling the "drivebar" in the "View" menu of the explorer allowed me to browse better (Explorer++ isn't looking very good in ROS yet). I don't know if it's just an explorer issue or something deeper (since my disk is ntfs and ROS can't read usb drives...).

Don't know if it's very helpful but since it was mentioned I figured it was worth the input.

[PurpleGurl]

I've been getting the same explorer issues with livecd builds from over a week or more... Though I thought it was related with me running it in MobaLiveCD (a small self contained qemu vm app... I like for it's size portability and ease of use even though it lacks configuration).

I found that toggling the "drivebar" in the "View" menu of the explorer allowed me to browse better (Explorer++ isn't looking very good in ROS yet). I don't know if it's just an explorer issue or something deeper (since my disk is ntfs and ROS can't read usb drives...).

Don't know if it's very helpful but since it was mentioned I figured it was worth the input.

Interesting. What was the oldest revision you ran with the explorer problems? I ran it on real hardware, not in a VM. I guess I should run it again and see if I can give the exception. I've searched Bugzilla and this explorer regression is not in there. So I probably should report it if nobody else does.

I am sure what you said is helpful, since any lead we can give the developers is a start. Things might not seem like much, but when put with what others notice, it may be enough to steer developers in the correct direction.

I would love to be able to try the one game I used as a benchmark for progress. The menus caused a crash and the left arrow didn't work, while the others got stuck. I've seen reports of fixing memory leaks/corruption and keyboard code changes, so we can see if there are changes there. But the explorer crashing won't let me test.

[frmariam]

I successively divided the release numbers in half between your test and August to decrease the number of tests needed in determining the culprit.
The issue seems to have been introduced in revision 53653 if I got it right (there is a gap between this and 53649)... Hope you can confirm it.

This one will surely be solved before 0.3.14 (can't imagine a release with a broken explorer).

[PurpleGurl]

I successively divided the release numbers in half between your test and August to decrease the number of tests needed in determining the culprit.
The issue seems to have been introduced in revision 53653 if I got it right (there is a gap between this and 53649)... Hope you can confirm it.

This one will surely be solved before 0.3.14 (can't imagine a release with a broken explorer).

Wow! This is a break. So it could be Shell32 related? Presumably, the missing numbers are portions that were in branches at the time. So I guess I should burn 53649 and 53643 and get back to you.

[SHELL32]
* Reintegrate the c++ shell32 branch. Exemplary team-work.. kudos !
* Better code quality, more tests run with less failures... and more.
* Dedicated to everyone who helped

This could be where the regression entered, since it is apparently a prerequisite, and it was an extensive rewrite of about 227 changes.

[Haos]

As you can see here: bug 6536
there is a dedicated metabug for all regressions introduced by new shell32. It has been set with 0.3.14 milestone, hence we should expect those to be fixed for the nearest release.

[PurpleGurl]

As you can see here: bug 6536
there is a dedicated metabug for all regressions introduced by new shell32. It has been set with 0.3.14 milestone, hence we should expect those to be fixed for the nearest release.

So what should we do? Discuss our findings under the metabug or file a new bug and link it back to it? I'll get the 2 trunks and see if I can confirm the breakage, and wait for additional instructions.

Okay, I tried both revisions, 53649 and 53653. 53653 was actually worse than 53952 because not only were the icons missing, and not only did the explorer give exceptions, you also couldn't directly open a folder because Reactos complained that the program was missing. Then when you use the browse files option, you then get the exceptions. However, in 53649, the icons appeared on the desktop, and there were no explorer exception errors.

I might as well mention other things of note regarding 53649. The mouse is as stable as it is currently, which is pretty good. I noticed I was taken wrong above when I mentioned the mouse in 53952, since I was saying it was working better. In older ones, the mouse would act funny during 2nd stage, but work fine in 3rd. Now, it either doesn't work at all until stage 3 (race condition?), or it moves normally in stage 2. So that was actually a compliment, not a complaint. Anyway, I tried the Gems 2.11 Sega emulator and loaded Traysia. I noticed that in 53649, the keyboard problems were there in the game. The left arrow didn't work, the other arrows would stick (the left would stop the sticky buttons but do nothing else), and the num lock did nothing. (bug 4405) As far as menus go, that seems to be fixed. In older versions, repeatedly opening menus would cause a crash. But the memory leak or buffer overrun seems to have been fixed.

[frmariam]

The error explorer gives me when I press "Explore" in the start menu is:

"ShellClasses exception

unknown Exception: 0x80004005
Context: ShellItemEnumerator::ShellItemEnumerator ()
Location: base/shell/explorer/utilities/shellclasses.h:1051"

[vicmarcal]

The error explorer gives me when I press "Explore" in the start menu is:

"ShellClasses exception

unknown Exception: 0x80004005
Context: ShellItemEnumerator::ShellItemEnumerator ()
Location: base/shell/explorer/utilities/shellclasses.h:1051"

Have you checked if it is already reported?if not, please check and report in case we aren't aware of it.
It could be linked to the new shell32 rewrite.

[Haos]

So what should we do? Discuss our findings under the metabug or file a new bug and link it back to it? I'll get the 2 trunks and see if I can confirm the breakage, and wait for additional instructions.

No. Metabug is one of the special tags we apply to bug summaries to cathegorise them. There are two kinds of tags, main and auxilary. Main tags are: METABUG, REGRESSION, PATCH, TRANSLATION, HACK, WIP (work in progress), UNIMPLEMENTED, whereas auxilary point to application name or specific module, like: win32k, explorer, shell32, Firefox, 7zip.


Answering your question on what to do: the idea behind Metabug is to list bugs that are similar (keyboard issues), originate from certain change|revision (ldr/shell rewrite metabugs) or come from same module (UNIATA metabug). We simply have specific bugs set to block the metabug. Such list is easier to maintain and can be accessed very quickly. You dont need to do anything with the Metabug itself, this is maintainer's work, only the specific bug reports.

For more info please check the follwing wiki sites:
http://www.reactos.org/wiki/Buglist
http://www.reactos.org/wiki/Bug_Filters

[PurpleGurl]

Have you checked if it is already reported?if not, please check and report in case we aren't aware of it.
It could be linked to the new shell32 rewrite.

Just added:
http://www.reactos.org/bugzilla/show_bug.cgi?id=6541

I hope I did it right. If I erred in how I do it, I ask for patience. Oh, I see I goofed. For alias, I assumed my alias. I don't think I was supposed to put anything there. Okay, just figured out how to fix that.

---
Bug still present as of revision 54059.