JRD wrote:We can try to build another ISO using grub2 directly (as before) and using isohybrid with grub2 directly.
@
JRD,
The computers mentioned by
mimosa seem to be different enough between each other, and yet in qemu the CD boots in those systems. So, instead of changing the current booting methods, maybe a simpler test by
mimosa could help? Using CD-RW would allow testing without having to throw out CDs.
Anyway, my proposal would be to expand the current MATE ISO, delete "[BOOT]", replace isolinux.bin with the one from Syslinux v.3.84, re-build a new ISO, test it in qemu and if that works OK then burn it to a CD-RW. With this new burnt ISO, test all 3 systems again.
This new ISO should not use isohybrid, and should use the same old version of isolinux.bin that was used for SalixLive Xfce, and
it would be very helpful to post the EXACT mkisofs command line with ALL the relevant parameters that should be used.
Of course the BIOS settings and specific hardware could still be related (
which exact settings we still don't know), but if changing the isolinux version works, then we can narrow down the potential solutions to be related to either:
A_ isolinux regression bug; or
B_ isolinux retired feature; or
C_ isohybrid issue; or
D_ isolinux.cfg commands that may need small changes for wider compatibility.
Of course all this would depend on
mimosa and
JRD, and providing all the relevant info before and after the tests, so this is just a suggestion with little actual collaboration from my part
.
My point is that whichever change is made for future ISO images, we still don't really know escential info and where the problem resides.
[EDIT]
@
JRD,
Remember that the systems are booting correctly from USB, and "from CD" when the USB is connected, stopping at grub2 (after isolinux), so knowing more about the hardware and the BIOS settings is important, for whichever change you would like to try.
[/EDIT]