MTR 6 ANNOUNCEMENTS - READ FIRST!

[geezerbuttz]

been home for a few hours...


Fixed the email iisue: Back ports to older system on < Os11
Fixed the restart issue , found a method only available on 10.11 ( why xcode gave me no warnings as my base is for 10:10 i have no idea ..

fixed the mkvMerge download crash..... was a problem posting a NSVisualEffectView. 11 onward is a lot more freindly on views that may be differed.

Tomorrow I will go through the light mode version of the app and tweak the design colors a bit , ... it looks somewhat lame while dark mode shines brightly ( no pun intended )

will test again on 10:10,11,12,13,14,15,11.. and rest of machines here to just be sure..
keep the dial tuned....

[geezerbuttz]

Well , it looks like 10.X is simply not good enough to handle the complex interface I designed.
while it runs on 10:10 , at this level , the code behind the UI is not powerful enough to give a good user experience.

10:13 seems better , so I will make 10:13 its minimum requirement.

from 10:14 onwards it will always run in darkmode for the best look... It was designed with this in mind.

Still installing to make some final checks...

I i need to add an extra note here: I am running these virtually , so you will get a better experience than Im getting on 100% hardware.. I will start to verify this tomorrow... )

[geezerbuttz]

Yes , confirmed 10.14 onwards looks and behaves..

I will do some real installations on a spare volume tomorrow to do some final checks..
I will check for 10.12 -> 10.15 just to be sure...
G

[geezerbuttz]

I listened , so i have been working on cleaning the Ui code for the older OS as well as the newer.

It is now fully usable in 10.10:11:12:13
still have to check 10.14:15 tomorrow.

Cleaned a lot of the UI up.

Installing virtual for test , and 100% HD install for final.

Time consuming...

[geezerbuttz]

Its going well... had a hickup timewise today , but progressing:

The UI update , vastly improves the overall look , even in darkmode.
Getting a bit more separation in the tableviews greatly helps reduce clutter..

I have just completed 10.14 with no issues , now on to 10.15 , then 13.X then 15.X ( i do not have 14 to test anylonger )

Getting there slowly , I do need to be confident

[geezerbuttz]

FYI , for those who are interested the testing plan is as follows.

Take a nasty disk.
Run it through SectorProbe to generate a map , run it through SectorScan to generate a map, Extract full disk.

This way I test all functionality in place.

[geezerbuttz]

found a bug last night related to the older OS.

When reading from a NSURLSession , the retuned response changes dependingif its a http request , or local file request.
On the newer OS , its a lot more relaxed and does not error out , older , not so.

A quick rethink not to use the statusCode method , and we are good again...

Just an update.

[geezerbuttz]

ok , caugh an issue in 10.10.
Done.

I had one crash in 10.10 but have not been able to reproduce it. Im suspecting its cased by the virtual machine not handling the disk correctly.
So , tomorrow i will install 10.10 for real to be double sure.

So upto now:
10.11
10.12
10.13
10.14
10.15
11.xx
12.xx
13.xx
14.xx
15.xx

all confirmed , license , activated , maps generated , extracted.

[TheSystemHasFailed]

Wow.. that's an incredible level of backward compatibility. How many machines do you run builds against?

[geezerbuttz]

4 machines , installing what I can , and what will still install as the needs arise...

Need to raise some funds from this new release to get a new studio to replace ageing iMac...
Its rather long in the tooth.

[TheSystemHasFailed]

4 machines , installing what I can , and what will still install as the needs arise...

Need to raise some funds from this new release to get a new studio to replace ageing iMac...
Its rather long in the tooth.

Might I suggest the new Mac Mini M4 Pro.. I bought that instead of the Studio.. the amount of horsepower for the price tag was insane.. This will likely be the last computer I buy before I retire..

Or go into business with you... I mean it's not impossible.. just improbable..

[geezerbuttz]

i looked at the mini... its ok , i just need the extra connectivity. But ...

who knows what the new studio with M4 will be like.
G

[TheSystemHasFailed]

who knows what the new studio with M4 will be like.
G

I can only imagine, brother... When I got the Apple //e in High School I hacked with that thing all the time until college and CP/M became harder to use for programming assignments... then built my own PC... and continued to use PCs for many years (skipping the whole M68000 and PowerPC Mac eras).

I was stunned at how much power the 2006 Mac Pro (my return to Apple after dealing with that horror that was Vista)... (I mean I think it was more powerful than the Cray we had at my university...)

Today I cannot even think of using anything but Mac and they just get more and more insanely powerful...

[geezerbuttz]

amen to that..

You know its not the hardware that is important , hackintoshes have proven that.
Its the Operating system... the whole infrastructure that we have is so much more friendly compared to windows...

[geezerbuttz]

Sad to announce:

The Demise of OS 10.10.

Alas , he was simply not able to perform in a way that would render him useful.

On a more serious note:
The renderer in 10.10 is not good enough to render the tableview elements cleanly when you have tableviews stacked in tableviews.

So rather than compromise , i made the choice to push the minimum spec up a notch to 10.11.

Sorry if that effects anybody.

[geezerbuttz]

ok in shifting to 10.11 for minimum OS meant i had to clean up some depreciated code.
It was good it made me update the downloading mechanism for external files.

MKVDownloader got a complete rewrite , its now flexible for re use across multiple classes.
You can instantiate it and feed it a block to hand the downloaded result... so it is flexible...
( the block only needs the files location on the system , and just to make things a little less complicated , a pointer to the downloader class , makes it easy to populate the blocks inner workings )

The download UI window opens on the key window , centered , but can be relocated as needs....

Reduced the older code....
from 5 different classes with multiple delegates , to just one.

Much better...Need to verify tomorrow that nothing got broken along the way...
Thats it.

[geezerbuttz]

well that was short lived.

I took the complete download code and wrapped it in a framework.

Added another block ( nullable ) that you can add any presentation code in ( i call my notification code , you could pop a window on the screen if you like... it only takes one param , and thats the message you want to show.

It grabs the hosting apps icon for its display panel when downloading...

If the message block is NULL , then no messages get displayed.

Much better to do it this way , reusable code , we love it.

You know blocks in frameworks are super cool , you can change the inner workings of a method when ever you call the framework.... As long as the block you provide matched the blocks prototype , its all good.....

Testing tomorrow this new code is ( said in a yodaesque voice )

[geezerbuttz]

Finalised the download addition of the LInuxPack.
Its linked into the download section of the Main Menu Bar with mkvmerge.

A small change that bothered me.

Sheets.....

On the older OS the sheet drops down from the menubar , on the newer it presents as a panel in the center of the window.
That looks nicense , so , i implemented a version for < OS11 that we now use.

It does not have the animated bells and whistles , but it looks far tidier.

Thats about it.... im packaging tomorrow.

[geezerbuttz]

Last checks:

10.11 , 10.12 , 13,14 , bug Sur ( 11 )( did not bother with 10.15 ... if 10.14 is ok as is 11 , it stands to reason 10.15 is ok )

all ok...

moving on in the morning , as the next are all real hardware...

[geezerbuttz]

Just completed qualification on OS13, and 14.

Made an addition and a quick tune in code.

Preview played back so fast that the drive could not keep up with it when playing from disk , leading to a constant stutter in playback.

I cured this by introducing a new preference in PostProcess / Preview prefs to control the playback speed.

X1,X2,X4.( can be changed at any time , default is X2 )

Re set the buffer to minimise memory footprint for small systems.

On to Os15 in the morning...

Its all working well.

[TheSystemHasFailed]

I'm sure I speak for everyone here when I say I cannot WAIT to try the new version out... I'll be warming up my paypal account to send another gift....

[geezerbuttz]

Ok , its 02:30.

Just completed running it through the profiler.

Found a few memory leaks , and some late deallocations. All taken care of.
Fixed a window miss-placement prior to 11.

Updated Preview code , removing dependency of main feature mode.
I will move it into a framework later , and add the ness hooks to use it from ISO mode.
makes sense.
All framerate , aspect ratio , cell , vob info is read from the stream.
Improved the memory allocation further.
Updated to include x8 and x16 fast forward.
Oh , i moved the speed popup onto the playback panel , away from prefs... it makes sense.

I was hoping to get this packaged today , but profiling and clearing the last issues to ok all day..
I hope tomorrow.

G

[geezerbuttz]

A lesson for all:

Chasing ones tail is never much fun.

Let me explane.
NSTableViews. Table view use cells or views. I like to use views as it makes it visually nicer to interact with.
It looked like I has a memory leak in my tableview code , I though I was not releasing the underlying data structures associated with the table view on time , when i removed the view.

In fact I kept seeing memory use go up.

I did notice that if i load one View , remove it , load another , the memory use stayed the same.
But , if i load one view , and then add another view the memory use doubles.... and so on.

The result:
TableViews cache the views. I always thought that it could reuse and existing view , but failed to understand NOT a view already being displayed.

Table views cache the views ( yes I said it twice ) , it caches ALL new views that get created and removed... This leads to and ever increasing memory footprint every time you create a new view ( never before used ) on the table view.

So , it not a leak , its not an over allocations , its tableview caching ints view data........

tale caught , back to the grind of wrapping this up.

[geezerbuttz]

Not for wanting , but i found a patch of incomplete code that stopped dvd ifo rebuilding... done.
changed the whole main feature UI code to reduce memory usage even more.

Found a few race conditions that I needed to take care of , a few semaphores did the trick...

Last but not least , mkisofs did not work on older systems , so we now include 2 builds and switch as needed....
Im sure there is mor , but i turned 60 today , so im forgetting things lol....
Phew...

Im sure there is mor , but i turned 60 today , so im forgetting things lol....

Happy Birthday!

[rthpjm]

Not for wanting , but i found a patch of incomplete code that stopped dvd ifo rebuilding... done.
changed the whole main feature UI code to reduce memory usage even more.

Found a few race conditions that I needed to take care of , a few semaphores did the trick...

Last but not least , mkisofs did not work on older systems , so we now include 2 builds and switch as needed....
Im sure there is mor , but i turned 60 today , so im forgetting things lol....
Phew...

Happy birthday GB

[geezerbuttz]

ok , i got the NSTask sessions running now under 10.11 -> 15

Sorted some core animation transaction issus...

Back to testing the updates....

[w3ato]

Im sure there is mor , but i turned 60 today , so im forgetting things lol....

Your catching up to me!

Happy Birthday

[geezerbuttz]

if anybody is interested:

The failure in OS10.11 that i encountered is as such.

I was doing view updates onto the main thread , but onto a queue. Normaly , if the view hosting those controls is hanging around , this is ok..

But in this case the hosting view is a tableView view, and that gets removed at the end of process. This can lead to view updates still being queued on the main threads queue that no longer have any reference to an active control on an active view.

That is the transaction error.

Rule of thumb , for tableView views always push onto the main thread.

As for now , im soak testing the new code by throwing every possible combination at it , and so far in 10.11 its running great.

[geezerbuttz]

resolving a potential issue.

When working fast I use ISO's.
ISOs can have padded space in them , blank areas. This caused MTR to have a little issue .
( no data for me to rebuild )
Its good I found this , its one more thing i dont have to worry about later.

[geezerbuttz]

as from new , post process operations have a lower priority than main and save process threads.
Sp , the post process takes a bit longer , but main and save , run at full speed ..

A better balance of cpu and disk usage..

Corrected a predicate error in a NSCondition logic.

Od course im reading from three ISO from the same volume and writing the results back to the same location , so im really hammering the IO.

Its all in stress testing...

If you are reading from three different drives to one location , the IO is far les stressed.