Syslinux - Aug 2015 - [PATCH 2/2] core/graphics: fix lss16 parsing

On Sat, 2015-08-22 at 00:56 -0700, Patrick Masotta
wrote:
> Please take a moment and see what the quoted code does; it'll save you
lot of time.
I wrote my versions before I knew about the other code.  Unfortunately,
that other code is a massive single patch and would be hard to review
since it would be difficult to bisect should a bug be introduced.  I
will try to split the other patch into logical parts and submit as time permits.

>>>
> Please take a moment and see what the quoted code does; it'll save you
lot of time.
 
 I wrote my versions before I knew about the other code.? 
<<<

Yes, but you published your patches after you knew about it.

>>>
 Unfortunately,
 that other code is a massive single patch and would be hard to review
 since it would be difficult to bisect should a bug be introduced.? I
 will try to split the other patch into logical parts and submit as time
 permits.
 <<<

Yes I know; unfortunately solving the lss16 issue cannot be done with a couple
of 4 line patches.
The code to fix was originally ported to C from the assembler of a previous SL
version, and it
never really worked before; it was needed re-writing certain parts.
Now the new code ( http://www.syslinux.org/archives/2015-July/023835.html ) has
been
tested and it does work (you can see an lss16 image in your terminal). I always
prefer start
fixing things from a working code base than start patching one that never did.

if you want to continue with the task probably it's more valuable if you try
to tackle
the associated issues related to terminal/cli handling rather that splitting the
big patch.
just my 2 cents.

Best,
Patrick

On 08/23/2015 12:33 AM, Patrick Masotta via Syslinux
wrote:
> 
>>>>
>  Unfortunately,
>  that other code is a massive single patch and would be hard to review
>  since it would be difficult to bisect should a bug be introduced.  I
>  will try to split the other patch into logical parts and submit as time
>  permits.
>  <<<
> 
> Yes I know; unfortunately solving the lss16 issue cannot be done with a
couple of 4 line patches.
> The code to fix was originally ported to C from the assembler of a previous
SL version, and it
> never really worked before; it was needed re-writing certain parts.
> Now the new code ( http://www.syslinux.org/archives/2015-July/023835.html )
has been
> tested and it does work (you can see an lss16 image in your terminal). I
always prefer start
> fixing things from a working code base than start patching one that never
did.
> 
> if you want to continue with the task probably it's more valuable if
you try to tackle
> the associated issues related to terminal/cli handling rather that
splitting the big patch.
> just my 2 cents.
> 
I do want to emphasize what I have already said, which is that I would
prefer for this to be done via the VESA subsystem rather than doing
low-level hacking with the VGA display.  One reason is that the quality
of video BIOS continues to drop, and we see more and more bugs in text
output on a graphics screen since no modern OS uses it, and for non-BIOS
it doesn't work at all.  Note that the VESA subsystem can already decode
LSS16.  However, at the same time it is obviously clear that users can't
wait forever.

	-hpa