MTR 6 ANNOUNCEMENTS - READ FIRST!

[geezerbuttz]

well , the packer side is working , sort off..
Still need to put in some error handling.

Takes an ISO , breaks it down into random size chunks , encrypts each chunk with a key that mutates every time its is used.
( every chunk created has a different key , evaluated from a master key set at start )

Each chunk is then compressed.

Each chunk is then added into a .tar archive,

The tar archive is not password protected , as there is no need.
( the reason for a tar archive is that it is the fastest to append to , without creating a secondary file that we add to , increasing the time taken exponentially )

oh yes , the contents are extracted in a random order , making it nearly impossible to reassemble outside of the app , without a correct password.

Just a little fun for those who need to provide ISO for education and evaluation use.

[geezerbuttz]

an added bonus , i have hashed every chunk to make sure they are not manipulated.
We have an encrypted TOC at the beginning that we use to verify everything.

The TOC is also encrypted , and contains a signature to verify its integrity.

New year tomorrow , health and happiness to all of you, i hope 2025 will be kinder.

G

[geezerbuttz]

right , fun day..

to much enjoyment last night

Well , i decided to not use ant external packers (zip:tar etc ) on the ISOPacker code, as im encrypting and packing the chunks anyway
all i need to do is output the stream in in blocks...

But to do that with need a custom TOC.
Thats done , generated , and will be written first , followed by the ISO data....

Oh , the TOC is encrypted, and compressed.....
Good luck reading that.

More soon from the funny farm.

[geezerbuttz]

packer is done , need to clean up the code...

No more external tasks...

disk is broken into packs , packs are hashed to enable verify on rebuild.
each pack is then encrypted , and the compressed... you dont get alot of compression on a stream as its already compressed video.

we then write out the stream , followed by the TOC.

We use the master password ( user choice ) that matches the encoding pw to decrypt and restore the image.
( that is why we need a TOC )

verified 100%
G

[geezerbuttz]

A small amendment to the packer.
The root table that pointed to the TOC was exposed, so i reworked the header to pack and encrypt that data.

That is now secure.

There is only 16bits of data that are unencrypted.

It will be quite hard to identify what is in the pack , if impossible....

[geezerbuttz]

So now on to the preview for ISO mode.

[geezerbuttz]

Ok ,.
Preview for ISO is now enabled.

I tuned the code to make sure that IFO,BUP Lead IN Lead OUT or not selected for preview.
Void Packs can be played back , giving you the chance to see what is being left out , and allowing you to decide wether to add them back in.

If you choose a void pack, or damaged pack with corrupted data , the normal rules apply , you will need to wait for the drive to relinquish control of the disk before the preview exits.

Due to the fact that menus can be viewed , I have added a 1 second delay to the end of playback so you can catch the content of the short menus.

This 1 second delay is not in effect if you choose to cancel playback.

the preview button is embedded in the ISO Cell Slice.

Thats all for the first point update.
( im just waiting on a user to report his issues with his mac that could not extract anything, the machine refuses to give him HD/or OD access it seems...... Neither V5 or V6 work )

[geezerbuttz]

Preview does not respect multi angle streams..

Im on it.

updated the code to import DVDSS and IMF into MTR , forgot in one case to handle angle blocks correctly.
Improved handling of partial blocks.

G

[geezerbuttz]

Thats preview re written to follow ILVU pointers in MF mode.
In ISO , as the streams are broken down into blocks anyway , its is not really going to be able to do that...
Its just how the ISO is structured..

Still , we have full preview capability now....

[geezerbuttz]

Plan tomorrow to implement a filter on the loaded cells to remove outliers that sometimes come from the IMF/DVDCSS files...
G

[geezerbuttz]

outlier filter is done , you can bypass it in user prefs/Batch/EnableOutlier rejection.

Removed all MTR dependent code from from preview , and wrapped a framework around it.
Works well.

Added an extra check to the IMF loader to prevent missing cells on a non sequential map.

Tomorrow:
Have a look over the standard cell selector in MF mode , as it is not always picking a good choice in cells.
Add preview layer to DVDSectorProbe.

Finecomb the data for redundancies...
G

[geezerbuttz]

Slight change of things done today.

Updated to fix a few issues.

Redundant sets were stripped before rootSector calculation was done. This has been changed to prevent missing root sectors in the table that are needed for the Main Feature engine.
( title that are displayed on obfuscated disks need the root sector )

ISO Cell Slices were calculated wrongly. It a rare case , a void entry start sector ended up at 1 , while the end was set normally.
The effect of this was void packs getting larger the further the extraction went. ( 20 hrs Abner )
That is now fixed.

End sector calculation was overlapping the last valid cell area.
Fixed.

Duplicate title removal no longer worked.
Fixed.
( a change of operation as follows:
Once a title is loaded , the dupes are automatically taken care of. On a IMF/DVDSS load , removal is NOT automatic. This is incase you want to peruse the raw structure of titles. You need to use the right click contextual menu to manually select the function. To reset back to raw state ( pre IMF etc ) use the reset menu in the afore mentioned menu )

Thats it!

Continue tomorrow.

[geezerbuttz]

Found out last night that MF UI was not working correctly.
When changing disks in the table the titles panel would reset itself , removing the IMF/reduction slections.

Reworked all to reduce memory overheads , and corrected error.
G

[geezerbuttz]

Profiled the new UI code , no memory leaks... at all..
Im quite amazed as its rather a complex structure going on there...


Thats DVDSectorProbe updated with , a few tweak and the addition of the preview code.
,VOB and .DAT can be previewed no need to play an IFO lol...
( version bumped to 1.1 , output is the same , so all old maps are valid )

That leaves this weekend to sort the initial cell entry / exit code that seems un reliable.

I am thinking to update sector scan with the option to preview the table of data ( will look like probe layout ) at the end of the session..

Need to think about that more....
G

[geezerbuttz]

Well , that required quite bit of work.

DVDSectorScan originated on Linux , and was ported to MacOS.
Therefore a lot of the goodness we have was missing , and was just added.

I compressed the UI in the vertical axis a little , added a table view , got it to eventually bind to my array controller.
Then addressed the color identification system , and it layout.

Lastly added the preview framework to the app , and added the UI hooks.

It all works.

One point , having three apps that now all share the preview framework , im thinking a of rebuilding it so it put into the frameworks dir in ~/Library.
that way it is one less thing to bundle together in updates as we link form the framework at run time.

Thinking about it.

Tomorrow , back on track with the initial cell entry code that is ropey.
G

[geezerbuttz]

That is enough for today:

Completed the move to the new colour pallet for the plugins , works in Light and Dark Mode.

I will get it into MTR at some point...

[geezerbuttz]

A lot more testing:

Currently in sector scan, and found a case that slowed it down badly. Some disks contain large areas of Navigation packs that simply point to the next pack in the stream... i.e , the very next pack.

This slowed us down as we were generating new slices for each single pack.
I fixed this by identifying the case , and makeing sure that we append these packs t the currently set video slice.

I also realised that this could be adapted and moved into MTR as an alternative ISO extraction engine.

[geezerbuttz]

Morning all...

Spent this morning looking at the sector probe disk layout result.
Not quite happy with the organisation of it.

I updated the code to calculate the correct and current VTS , augmented the UI to provide a list on sets that are duplicated in the FAT.

It looks like a popup , but its just a list.

If there is no duplication of a sector range , there is no list. Its simple.
It is there to guide you , and MTR , on what we are dealing with.

I plan to augment the output file format , and add the dupe mapped table to it.
I will bumt the file format version , so we can filter the difference.

[geezerbuttz]

completed the sectorscan update.

After a lot of deliberation and forward thinking , I added all the possible parts we need for the future.
every slice has its CellNmr , TitleNmr , VTSNmr , PGCnmr , And sector data , pack type etc built into it.... . That is only available in V2 of the file file format.

I now need to update MTR's title extractor to use the version 2 file format if it is available , as this will greatly simplify
The selection in the main feature table , as we have all the cells and titles now provided and no longer have to walk sector regions.


I will do that this week with a look to get the update out by the end of the week.
G

[geezerbuttz]

ok DVDSS import is done. The new import has an aux function.
Once you have imported you can use the "remove empty titles " menu item in the contextual menu.
This remove titles that have NO entries selected in the cell table.

The DVDSS makes a secondary check to verify the start sector of its cell does indeed lay on the correct disk boundary.
If not , it turns off the cell.

One disk I have only leaved 2 valid titles after these process , thus greatly reducing the choice you have for MF.

This is only for DVDSS imported sessions currently. I do not think that it is ness for IMF sessions.
G

[geezerbuttz]

titles mode has a new flag in its titles display that gives you a visual clue as to the selected condition of cells in that title.
If the Cyan flag is active there are no cells selected in that title.

This is an extra visual clue that goes hand in hand with the remove empty titles contextual menu item.
(Extra: added the selected titles count in the title count flag for easy look up , you no longer need to access the contextual menu to access this )
We move on.

[geezerbuttz]

Hi.

I have been doing a little analysis if the main feature entry cell code.
It is not really working for me currently.

I need to update it to return an array of cell states , and not simply an entry cell number.
This will allow us to map entry and exit cells better.

Also it is rather elaborate in it calculation , and it can be simplified some what by utilising the existing captured data better.

I will commence this on Monday AM.

This will give you a bteer starting point if you choose not to import maps.

[geezerbuttz]

Well , that was a catch.
I moved the root sector calculation to before the table rebuild to make sure we grab all the roots for the titles sets.
This was done to make sire we do not loose reference to sets that are in duplicated set structures.

The issue here was the I was not rebuilding the sector tables , and we were referencing the nonsense on the disk.
Thus the start sectors in MF were all over the place.

That is taken care off.

[geezerbuttz]

last bits of 6001 just completed.

Updated the MF option right click panel to display local start / end sector ,
Added a discontinuity indication to show when the are jumps in the cell data.

Redrew the option panel to match the look of the rest of the app.

Just a few bits to help to make the correct choice in the title table.

Bit of testing now.

[geezerbuttz]

well , once again i ran into issues with these root sectors in title mode.
The roots generated as they are , are really tuned for ISO generation and full disk rebuilding .

So , the only solution was to code a separate root table for the title mode.
Thats done , seems to be ok , i will however give it a full test.

This issue will most likely cause issues in the original release in mainfeature mode.
Sorry about that.

[geezerbuttz]

As David Bowie said... "Im going through changes...."

So , working still on this root sector issue , I was never happy with the solution in place.
Some disk were ok , some were not.

The problem here is that one new piece of code exposes you to new possibilities , so you investigate them.
This has led to a root system that can calculate the roots of all sets , and if requested return the roots of sets that are illegal mapped over other sets.

So , a new preference is in place in the batch title prefs to display unmapped titles.
In the off position , titles whose set is mapped between other sets ( over lapping ) will not be show.
In the on position they will be , but , the root sector is most likely bogus as these titles belong in overlapping VTS.

Its a new option to play with that can reduce titles in the MF dramatically. Of course , if the sets are mapped over the exact sector range of another set ( clone sector data ) then its set is returned as valid , thus so are the titles.

This is far more flexible , and allowed me to get rid off a large portion of older code.

Now I can test...

[geezerbuttz]

on a foot note, i will add a contextual menu to enable removal of the unmapped sets as well , and find a way to do double duty in the flags table to indicate them.

[geezerbuttz]

i removed the preference for stripping unmapped titles.
I moved a new menu into the title table ( as mentioned in the previous message )

I also added a helper function there that allows you to call all the title filters in one go.

I updated the flags code.

The top three flags stay the same , but the last has multiple states:
Cyan , no selected cell , yellow , unmapped title , and magenta if both no cells and unmapped title are true.

The tooltip for that table have now been done in code to allow better formatting of the information.

That is something I will look at expanding onto other complex tables tomorrow.
G

[geezerbuttz]

As normal , i decided that the initial solution was not good enough.
So today i re wrote all of the flag handling code, moving it from a on-the-fly drawing routine to proper classes.
It just easier to tune the drawing routines , hence we now have a full 5 flags in the title table , with no flags doing double duty.

So far Main Feature title table , and cell table flags are done , with the ISO slated for tomorrow.
I will also complete the tooltip expansion.

[geezerbuttz]

completed the flags code .. had to redo it all again this morning.
The reason:

I tried out 'paintcode App' and I was not impressed.
The code generation is bloated and not very efficient.
For example: the cells flag generated by the paintcode app was over 200 lines drawing code alone. The hand written version , a full class , was a shave over 60 lines of optimised code.

I fitted three separate class objects into the same space it would off fitted one.

Not to say it does not have its place in UI prototyping , but just not for release code.

Been correcting some UI constraints to draw the main UI better.

Testing tomorrow at long last.