Version 0.9y-2 feedback

[PeterSt]

For the Coverart Out of Memory Issue : Please try below XXHighend.exe. Paste it over your current XX folder and let me know when it helps (it should ).

Edit : Forgot, this solves the doubleclick issue on the Coverart as well.

[boleary]

Am off to work. Will run the tests you requested either tonight or tomorrow morning. Thanks!

[PeterSt]

boleary, thank you very much. And I am very sorry it goes like this. If I only had it myself ...

To be ahead of things, please also let me know the length (in time) of both files in question, so I can try to copy that too. It might be related and it may be a more normal bug on my side.

Have a nice day of work, and save a few.

[Calibrator]

For the Coverart Out of Memory Issue : Please try below XXHighend.exe. Paste it over your current XX folder and let me know when it helps (it should ).

Edit : Forgot, this solves the doubleclick issue on the Coverart as well.

Hey Peter,

thanks for the quick fix.

The out of memory issues and double click issue appear to be resolved now, however I am still having difficulties with the 'B. B. King' folder not showing folder art as I outlined above. Will continue tinkering to see if I can see what is common amongst any more ( if they are evident ) that don't show up.

Cheers,

Russ

[Calibrator]

Here's another bug for you to look into.

When attempting to play, in UN attended,  "Cannonball Adderley - Something Else" encoded in 24_96 flac, after the flac's had gone through their initial conversion process and the GUI dropped out of sight, I was presented with a blank screen ( was expecting the cover art next ) and about 5 seconds afterwards an Out Of Memory error box appeared. I clicked OK and nothing further happened. Alt-X to get back to GUI was the action.

Initially I had the split file size setting at 100MB then bumped it up to 128MB but it made no difference.

Tried it a couple of times with same result, so turned logging on for last attempt.

Tried again in ATTENDED mode and same results.

Logs attached ( hopefully in chronological order )

  soon for me.

Russ

[Eric]

For the Coverart Out of Memory Issue : Please try below XXHighend.exe. Paste it over your current XX folder and let me know when it helps (it should ).

Edit : Forgot, this solves the doubleclick issue on the Coverart as well.

Peter,
this version 2a solves my Out Of Memory problem.
Thanks, Eric.

[PeterSt]

G'day Russ,

Well, I forgot about the B.B. King issue. Must be fairly easy to mimic - and solve I guess. Sorry I forgot.

Something Else ... I have that too (a most good recording for 1958 !), although in WAV, but the exact same thing happens.
Btw, to solve such a thing DEcrease the file parts, so they will be using less memory.

But now I have something ...

To me it looks like this happens more often (out of memory by normal means - not bugs like the Coverart thing) than before. Now, this is a nasty one, because of some parts running in a special area of code, that code indeed using less memory because it wasn't preprocessed (same like ticking the Mem checkbox from before). Now, since I don't recall that anything changed on the memory management (which is rough by itself because of the long period it took for 0.9y to emerge), I blame it to everything being preprocessed now, and nothing processed in real time. So, I have sure run this album before a couple of times, and indeed now it doesn't work anymore.
Must investigate this further.

Btw, if you DEcrease the file parts it will go eventually.

... Now I think of it, native 24/96 files indeed ran in that piece of code that processed in real time (only padding was needed). So, not anymore now, which immediately creates a problem for larger files like the first track here which is 370MB, will expand to 500MB because padding to 32 bits (in your and my case) and which is needed a couple of times in memory. Cutting that to 100MB still gives a net size (per instance) of 160MB (I just looked), which makes me think I don't get it.

.. hmm ... I just realized that I initialized an array larger than what is initially (!) needed when Split is set at 100MB (and where the Splitting came later (for development) than this initializing of arrays). So I lowered this (but it gets expanded as before) and now I use 1.2GB instead of 1.5GB earlier ?
The mysteries of Windows ...

I will see if anomalies come from this, and if not I'll share it with you (all).
As usual, Thanks !
Peter

[Calibrator]

Btw, if you DEcrease the file parts it will go eventually.

hey Peter ..

got the album to work in ATTENDED mode using a split size of 90MB, but when I attenpted to play UNattended it didnt work, so dropped the split size incrementally to 80 before it played. Thing is though, it wasnt consistent. Sometimes it would play UNattened at 80MB settings and other times it wouldnt. This was even after a clean restart of XXHE each time using the same setting.

I shall await your magic

Cheers,

Russ

[Calibrator]

A followup to the above:

Playing in unattended, it brought up the outofmemory exception message at the conclusion of the 2nd track.

Think I'll stick to standard 16/44.1 tracks for the time being 

Ta,

Russ

[PeterSt]

Hi Russ,

Was this perhaps still ok at the first 0.9y version(s) ?

(don't need to test if you don't know right now)

Peter

[Calibrator]

Hi Russ,

Was this perhaps still ok at the first 0.9y version(s) ?

(don't need to test if you don't know right now)

Peter

Just plonked 0.9y-1 back onto the HTPC for testing.

It is a lot more consistent in playing the Cannonball Adderley album. It started OK in UNattended at a 100MB split setting, so I gradually increased the setting by 1 MB each time, clearing the playlist and shutting down XXHE in between each test so it would have to rebuild the wav's each time.

At around the 125MB setting I started to see inconsistencies in actual sound startup. Sometime the coverart would appear but no sound happened, or sometimes it would start playing the first track fine. This inconsistent start is likely another glitch inherent in that early 'y' version, but at least I'm not seeing the OutOfMemory condition with that early version.

Haven't had a chance to play the album all the way through yet. I know the opening bars of track one off by heart now though .. haha

Hope that helps a little

Cheers,

Russ

[boleary]

Hi Peter, here are the numbers you requested when switching between 16/44 and 24/96 files, and vice versa. I don't think the DAC has anything to do with this issue because I ran the test once using the DAC and once just using the internal sound card in my laptop and got the exact same results. Here are the numbers:

Song 1, 16/44, length 3:26.42 to Song 2, 24/96, length 4:29.63 generates:
2/32/44100/352800/8 and 2/24/96000/576000/6

Song 2, 24/96, length 4:29.63 to Song 3, 16/44, length 5:45.04 generates:
2/32/9600/768000/8 and 2/24/44100/264600/6

Playing and stopping each song respectively generates:
Song 1: 2/32/44100/352800
Song 2: 2/32/96000/768000
Song 3: 2/32/44100/352800

Hope this helps.

[PeterSt]

First of all, thank you both for your efforts.

But oh dear, I start to be afraid that both issues are related, and my initial solutions for the "new memory management" as per 09w-5 from Dec. 18 2008 - which tweaks the untweakable dot-net memory management - are not working, or are not working anymore because of "something" that intervenes now. The memory-corrupting normal MS routine is a first proof of it (I think), the inconsistent behaviour is another and something working here but not at yours (boleary) is a next. And the worst is, while the corruption seems to be solved, other things are out of order now.

FYI, you can already see that my "tweaks" don't work anymore, because the memory can shrink again. This wasn't before, and it only could grow. The fact that it shrinks though, is an activity of the OS, which just needs it. But it can't need it, because there is sufficient in the first place.
This is back to before Dec. 2008, when this was the same. Or, back to the first 0.9y when things looked okay, but clearly were wrong. This (wrong) situation, however, may have cleared the unused memory by accident and could have been in there since always.

What can you do with this ? nothing. Nothing except this is not going to be easy to find/solve.
Btw Russ, I could run that track in 978MB (split at 110MB) which is just physical reality. Anything more is eaten by the OS, only freed when it reached the seeling (which it will eventually, ad how it was before Dec. 2008).

Russ, if I may ask, can you switch formats ? Thus, tick Allow Format Change, and let a 44.1 run into e.g. a 96 and back. Attended or Unattended doesn't matter much. Just the latest version (0.9y-2a). Maybe from this some equivalence or common denominator with either me or boleary emerges.

Thank you guys,
Peter

[JohanZ]

Sometimes at the start of first track (in unattended mode) the OSD Text is not shown. At the start of the second track the text is shown normally till the end of the playlist.

[Calibrator]

Russ, if I may ask, can you switch formats ? Thus, tick Allow Format Change, and let a 44.1 run into e.g. a 96 and back. Attended or Unattended doesn't matter much. Just the latest version (0.9y-2a). Maybe from this some equivalence or common denominator with either me or boleary emerges.

As requested ... using 0.9y-2a

In UNattended mode I queued up the first track of Cannonball Adderley, followed by the first track of an Acoustic Alchemy album (16/44.1), followed by the second track of the Adderley album , and finally the second track of the Alchemy album.

Without ticking the 'allow format change' button ( I wanted to see what would happen in that situation after the first track completed ), I pressed PLAY and it started up fine. I was called away to a phone call shortly afterwards and the screen saver kicked in. At the completion of the first track I was greeted by silence and when I moved the mouse to see what happened the GUI had popped back up and the second playlist entry was highlighted. If an error message 'had' popped up I didn't notice because I was away from the PC.

I then ticked the 'allow format change' box and pressed PLAY while the second track was highlighted. ( I don't think I shut XXHE down in between these sequences but the logs will show whether I did or not ). As hoped, the first of the Acoustic Alchemy tracks played till the end then the second of the Adderly tracks, followed by the second of the Alchemy tracks. So in this sutuation, ticking the 'allow format change' box seemed to work.

Ever curious to see what would happen if I started with the 24_96 track first, I then shut XXHE down, without clearing the play list, and fired it up again. The 4 entries were still in the playlist as expected and I pressed PLAY. Shortly after the GUI disappeared, and just as I was expecting to see the coverart appear, up pops a OutOfMemory Exception .... hmmmmm. Probably didn't like that 10 minute long 24/96 track.

Moved the logs for this test into a separate folder.

I also noted that yesterdays testing also had the logging turned on, so I've zipped all of those up also for you to look through in case anything is apparent, although I haven't got any running commentary as to what I was doing for those.

It sure has been an interesting way to spend one's birthday ... bug huntin' .. LOL

As always, happy to help.

Off to  for me soon.

Cheers,

Russ