-
Notifications
You must be signed in to change notification settings - Fork 250
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Missing Logs: Inconsistent and Unexpected I/O flush behavior #224
Comments
Friendly ping. I would appreciate review of this bug since I think it could affect future debugging ability with Proton 9.0. |
I've requested access. How are your running Proton? Are the lines missing form the log file created with I don't think I've ever had issues with Are you using software that you cannot modify that prints out things like that? Is using stdout/stderr necessary? TBH it sounds like a time sink that. may not be worth it if it's just for custom debugging programs when there are better mechanisms. |
I've run my test application both through the Steam client as an external "game" and through a Linux terminal with the short script below.
Yes. I modeled my test application to behave the same as the Paradox Launcher in ValveSoftware/Proton#7529 linked above and the game in that case did not have these prints in the proton log. ANGLE, used by the Paradox Launcher, was specifically using an fprintf to stderr, so I used printf to stderr in my test application. (Note from that bug that the mentions of
I applied
I wouldn't have had a way to modify the Paradox Launcher and I wouldn't have a way to modify any other game that might run into this.
Any game that leverages ANGLE at least has the possibility to printf to stderr since that is a pattern that ANGLE follows. There may be other software out there in use by games that may output informational messages in the same way. |
I've picked up Eric's upstream patches. Can you check with bleeding-edge?
Ah, thanks for the context! FWIW from my experience stdout/err is rarely used by games. Most engines log either to a file directly or use |
Unfortunately, I didn't find any difference after switching Proton Experimental to the bleeding-edge beta. The trick to unbind unix ttys still did not help. Also, I spot-checked a couple of tests from the spreadsheet I linked in the original bug description and didn't find any differences there either. |
Just checking in to see if there is anything I can help unblock in this issue. Please let me know. Thanks! |
Hi. It would be good to have understanding what's different between upstream Wine and Proton that causes the issue for you. Can you maybe compile standalone wine from our Proton tree and see how that behaves? If you can repro the issue maybe looking at our downstream patches. This would help to get us on the right track. |
Is "standalone wine" different from the wine binary produced by the Proton build flow? |
No. I've just meant that you can build Proton's Wine as a standalone and skip Few things that are IMO worth checking:
If 1. doesn't work it may be that we should backport more patches from upstream to at least be on par with it. |
Signed-off-by: Eric Pouech <epouech@codeweavers.com> (cherry picked from commit 417d25e) Link: ValveSoftware/wine#224
…reams. Signed-off-by: Eric Pouech <epouech@codeweavers.com> (cherry picked from commit d9b3ad0) Link: ValveSoftware/wine#224
Signed-off-by: Eric Pouech <epouech@codeweavers.com> (cherry picked from commit 8825f4d) Link: ValveSoftware/wine#224
This is a Proton Wine counterpart of an upstream Wine issue I just filed (summary below) describing how, under some conditions, I observed application prints missing from a Linux terminal and/or I/O redirect. This made my recent debugging of ValveSoftware/Proton#7529 much more challenging because I was working with limited information.
I found the behavior in Proton Experimental (9.0) to be substantially different (worse) from Wine 9.0 and 9.4 so I thought this should merit its own bug report.
In summary:
Full details here. I'm unable to make the drive fully public per my organization's policies so please request access.
The text was updated successfully, but these errors were encountered: