#wine-staging

that would be great yeah

just to maintain versioning

AmineKhaldi: yep

AmineKhaldi: btw, just curious, but has ROS also wow64 stuff implemented ?

I don't think so no

we're focusing on trunk (32bit)

due to limited resources

fork() failed: Resource temporarily unavailable

puk: your new pc? seems like the issue is more critical than you thought

now i need to bump up the system limits to have -j12 work correctly

nah

something in ulimit

no clue yet if its

max user processes (-u) 4096

open files (-n) 1024

\

puk:

michael@cyberman:~$ ulimit -u

63643

michael@cyberman:~$ ulimit -n

1024

yeah, it is the process limit

argh

lol

spawn -j$NR_PAR

and forgot to set $NR_PAR

this should result in a good multithreading test ;-)

it was fast

:)

and box is fine

but -j12 was fast enough too

without fork bombing the box

slackner: and the box boots lately

it even rebooted nicely once

puk: ah, already thought it could be related to the other issues you have with that box

when its just the software its possible to fix it of course ;)

any idea where to download small .avi file for a testing purpose?

AmineKhaldi: http://ix.io/gGa

AmineKhaldi: could you please test the patch (on top of 1.7.37)? if it works i am going to add the changes to wine-staging

sure, sec

AmineKhaldi: the struct in reactos also doesn't match exactly the docs, but i think it should be good enough

the ndk one ?

yes

slackner: EXCEPTION_REGISTRATION_RECORD

no Prev

AmineKhaldi: oh, more complication errors? :/

yeah

what does the reactos struct look like ?

ah, found it

lol, on reactos its "Next" instead of "Prev"

well so does http://www.nirsoft.net/kernel_struct/vista/EXCE...

and whatever that is: http://winappdbg.sourceforge.net/doc/v1.3/winap...

(Class Variable Details)

AmineKhaldi: wine policy prohibits us from looking at anything that might be MS-derived code

AmineKhaldi: i assume this part of wine is much older than any of the header files you mentioned above, either MS renamed it in the past

AmineKhaldi: or wine devs even reverse engineered it

ehoover: that's not only impractical, it's also impossible

because you can't do much without at least checking native headers

tell me how asm falls into stefan dosinger's mind then, when it comes to wined3d ?

:)

AmineKhaldi: headers are a gray area, but most of the time we can get that data from MSDN or from an application that uses the header

that's Ms derived in my book

MS*

it's kind of a game of "figure out how this works with both hands tied behind your back and one eye poked out"

AmineKhaldi: somehow MSDN and other apps don't count, you'd have to ask the lawyers

yet if you trace this app to see how windows interacts with it, it's a no

^^ i don't make the rules

i just follow them

we all do, but that doesn't mean we enjoy doing so

begrudgingly

very begrudgingly

"very reluctantly"

AmineKhaldi: lot of the code in this header file is from 1997, 1999, ... not really surprised that it doesn't match the definition of newer windows versions

AmineKhaldi: not sure if this can be fixed easily in wine, it affects a lot of files i think

AmineKhaldi: might be easier to make it an anonymous union in your header files

can I have the changes for just virtual.c ?

and apply them in ros

AmineKhaldi: sure, here are the test changes separately: http://ix.io/gGc

frame.Handler = guard_page_handler;

AmineKhaldi: whats the problem with that line? (btw, these are all no staging changes, you guys should sync more often *g* ;) )

virtual.c(1672) : error C2440: '=' : cannot convert from 'DWORD (__cdecl *) (EXCEPTION_RECORD ,EXCEPTION_REGISTRATION_RECORD ,CONTEXT ,EXCEPTION_REGISTRATION_RECORD **)' to 'PEXCEPTION_ROUTINE'

I changed all Prev to Next

but this field (Handler) has another structure type

another type*

AmineKhaldi: what directx version reactos is aiming?

AmineKhaldi: i assume another difference between the wine and reactos definitions, gcc would complain otherwise

hmm, 9 I guess, I'm not sure

slackner: oh it does

AmineKhaldi: i mean when compiling wine

I just favored msvc for the clear diagnosis

AmineKhaldi: can you show the definition for PEXCEPTION_ROUTINE

ah, yeah

the same structure

ok, sec

AmineKhaldi: thx, taking a look - btw, do you guys not use the ntdll tests ? they contain similar code since a very long time

hmm we do

let me see

no changes about exception stuff

but it uses winternl header

and wine/exception.h

this guard_page_handler look like it could simply be casted

well, except the NTAPI, which should be defined as STDCALL, right?

right

that means stack corruption

:)

can't guard_page_handler be set stdcall ?

AmineKhaldi: sure, but i am still wondering why wine has defined it different

but it seems like exception handlers have an integrated protection against stack corruption

"movl %ebp, %esp\n\t" /* restore saved stack, in case it was corrupted */

nevertheless, it should probably also be fixed in wine if the header files really say something different

I think I'll try to somehow use winternl until we fix all this

but for now you should be fine with casting and/or changing the declaration

(instead of using ndk...etc)

SECTION_BASIC_INFORMATION

is the first member pvoid or ulong ?

AmineKhaldi: pvoid

ok

phew! compiles

with winternl but still

Republique Remasted has no sound at all, tried with dsound and directx9 overrides. It uses unity5.

slackner, btw, if you have any GSOC ideas, please update the wiki (I see DarkPlayer already did)

of course, others can too :)

oh hey austin987 ;)

of course, it should be something that can go into upstream wine, not something wine-staging specific

howdy AmineKhaldi

austin987: well, in theory there is no "wine-staging specific task", upstream wine and wine-staging are both affected by the same bugs ;)

austin987: and under a different maintainer, a separate staging tree would have been never necessary ... ;)

austin987: but thats a different story ;) will think about some ideas and update the wiki soon

austin987, slackner: quartz cleanup ;)

ehoover: .... uuh, don't want to mentor such a task, lol

when's the mentor deadline? ;)

ehoover: afaik the list of ideas should already be "final", but its not really important anyway

ehoover: because students can also suggest their own ideas

eh, it's always good to add more ideas, but yes, we prefer students suggest their own