[AudioEngine] Support for AE on Windows platform

**jjd-uk** · 2012-08-08 13:34

The dev's working on AE put in a lot of work during the first 6 months of the year into getting AE merged into the XBMC master branch, although it can be frustrating from a end users point of view not seeing any updates please cut them some slack, as remember this is all done on freely on their own time and they are well deserving of a break after all their incredible efforts on AE thus far.

abudabi · 2012-08-10 17:58

Agreed, they deserve a break ... But....

They are just MIA and no one seems to know where / when they will be back. Or at least I gather that from the responses I have gotten. I think last time dddamian said he'd be gone for a month holiday, but he hasn't even signed in since he left and that was probably 2 months ago by now.

I don't think the community wants everything fixed now, I think it's just a case of knowing whats happening... or when to expect an update. No one likes to be kept in the dark.... And yes, the devs probably don't owe us anything, afterall they have done so much already... but it would be nice

Big Grin

abudabi · 2012-08-11 07:19

(2012-08-10 17:58)abudabi Wrote: but he hasn't even signed in since he left and that was probably 2 months ago by now.

Wahahahaha.. i just checked and he signed in 3 days ago. I give myself the Stare

Anyway.. at least he's still alive and kicking so lets see what happens. Rofl

**jjd-uk** · 2012-08-11 12:51

An update would be nice even if it's just a "haven't the time due to busy personal/work stuff" or "spent so much time on xbmc that wife threatened to divorce me so taking a break".

As AE is now merged into mainline branch then any of the dev's can work on it, but no doubt due to the other stuff like the new Android port then there's probably other priorities at the moment. There's also work needed I believe to get all platforms up to the same standard so maybe progression is halted on Windows builds until other platforms catch up.

What ever the reason for it being quiet around here this will definitely have to change once Team XBMC decide there is a need to get Frodo pushed out for an official release, so I would expect activity to pick up again towards end of year.

**thrak76** · 2012-08-11 13:08

I'm still using the Frodo Alpha 3 build, and really have had very little problems. AudioEngine works fine in that build. I'm patient enough to wait for work to resume on AudioEngine, and just like jjd-uk said, once the push moves toward the betas and then official release, all will be addressed.

Death-Axe · 2012-08-11 15:39

Would be nice to know what is going on indeed, and that issues like the fact that AAC isn't being processed correctly (audio discontinuity) are being looked into. I have been told that there is more than one dev working on AE but it would seem that non of them are looking at this thread at all.

**Martijn** · 2012-08-12 16:07

(2012-08-11 15:39)Death-Axe Wrote: Would be nice to know what is going on indeed, and that issues like the fact that AAC isn't being processed correctly (audio discontinuity) are being looked into. I have been told that there is more than one dev working on AE but it would seem that non of them are looking at this thread at all.

Maybe there are more important issues to be looked at atm.

ace20022 · 2012-08-12 16:41

Hi all,

since I'm also suffering from the freeze if playing music over a wasapi output I had a look at the code and the git log today.
The commit causing this issue seems to be https://github.com/xbmc/xbmc/commit/8ac5...ab24a092eb .
In particular the changes made in AESinkWASAPI.cpp.

I made a patch based on revision f2fd68ed03454885ca1b70c39f94cac51509d5ae : http://pastebin.com/7dtqAcnD .
It reverts GetDelay() and GetCacheTime() to the behavior before the above mentioned commit.

I hope this is useful for some of you.

ace

**davilla** · 2012-08-12 16:59

GetCacheTime better follow the API and return the amount of time that is buffered and not the amount of time remaining in the buffer before it fills. If not, then that patch is wrong.

ace20022 · 2012-08-12 17:22

I'm sure you are right! This is only ment to be a workaround.

As I'm totally unfamiliar with AE, I just reverted the code fragment to what it was before that particular commit.
Now the audio is played without any problems.

Before I reverted it, I got (sooner or later) an access violation at:

Code:

XBMC.exe!CAESinkWASAPI::AddPackets(unsigned char * data, unsigned int frames, bool hasAudio)  Line 441

Line 441: hr = m_pRenderClient->GetBuffer(NumFramesRequested, &buf);

in the CSoftAE thread,
while the PAPlayer thread is at:

Code:

XBMC.exe!CAESinkWASAPI::GetCacheTime()  Line 344

Line 344: HRESULT hr = m_pAudioClient->GetCurrentPadding(&numPaddingFrames);

abudabi Member Posts: 84 Joined: Aug 2011 Reputation: 1	2012-08-11 07:19 Post: #193 (2012-08-10 17:58)abudabi Wrote: but he hasn't even signed in since he left and that was probably 2 months ago by now. Wahahahaha.. i just checked and he signed in 3 days ago. I give myself the Anyway.. at least he's still alive and kicking so lets see what happens.
	find quote

davilla Team-XBMC Developer Posts: 10,404 Joined: Feb 2008 Reputation: 58	2012-08-12 16:59 Post: #199 GetCacheTime better follow the API and return the amount of time that is buffered and not the amount of time remaining in the buffer before it fills. If not, then that patch is wrong. MediaInfo : http://mediainfo.sourceforge.net/ Do not e-mail XBMC-Team members directly asking for support. Read/follow the forum rules.
	find quote