[Oberon] Project Oberon running from LPDDR memory on Pipistrello

[Michael Schierl]

Am 11.05.2020 um 14:37 schrieb Jörg:
>
>> 20180918 - considerably less heap space -- ORS.Mod.txt
>> What is the trick to use less heap?
>
> I guess it‘s a typo. Must be ORB iso ORS as the scanner doesn’t need heap space. If I rember correctly, some objects unintentionally got allocated more than once.

From the relevant diff on Spirit-of-Oberon GitHub I'd say it must be
this change in ORB:

<https://github.com/Spirit-of-Oberon/wirth-personal/commit/c288d035cbf655031c1e5a7123a81b6aa4e4157d#diff-ee9e5d88425b6bc686e78a7c14b65558R24-R28>

In RECORD ObjDesc, two INTEGER fields where changed to BYTE (class,
exno) and fields were reordered to reduce alignment overhead, saving 8
bytes in total. I did not verify, but probably this also means that
ObjDesc now fits in a smaller heap block than before.


Regards,


Michael