[Oberon] Oberon 2 compiler for RISC5

[Chris Burrows]

These RISC5 Oberon compiler RAM requirements pale into insignificance when
compared to an implementation of GNU Modula-2 compiler on Linux. My jaw just
dropped when I read an announcement on the Modula-2 Google groups about GNU
Modula-2 being available on a Beagle Board:

"I have finally gotten around to building a native GNU Modula-2 compiler
package for Debian Stretch on the BeagleBone.  It took about 9 hours to
build on a SanCloud BeagleBone Enhanced with 1 GB of RAM.  I wasn't able to
build successfully on a regular BeagleBone with only 512 MB of RAM."

Does anybody have any ideas why the GNU compiler is such a memory hog? If it
is the cost of portability I really think there has to be a better way (e.g.
the OP2 approach) of achieving that.

It hasn't got anything to do with the Modula-2 language itself. E.g. the
requirements for MacMETH, the ETHZ single-pass native code compiler for
Modula-2 that the original Oberon compiler was derived from, are:

"The MacMETH system requires at least a 512 kByte Macintosh and runs on all
Macintosh models up to the Quadra. For serious program development, 1 MB of
RAM and an external drive or preferably a hard disk is recommended." That's
more like it :)

Chris Burrows
CFB Software
http://www.astrobe.com/RISC5




> -----Original Message-----
> From: Oberon [mailto:oberon-bounces at lists.inf.ethz.ch] On Behalf Of
> Skulski, Wojciech
> Sent: Thursday, 8 November 2018 2:24 PM
> To: Michael McGaw; ETH Oberon and related systems
> Subject: Re: [Oberon] Oberon 2 compiler for RISC5
> 
> Michael:
> 
> both simple and OP2 compiler would be great to have. The RISC5 back
> end for the OP2 for cross compilation, and the native Oberon-2
> compiler for small embedded systems. At least I can see value in the
> full fledged FPGA-based Oberon System. I hope that Linz V4 can be
> hosted in four megabytes offered by RiskFive. A scaled down version
> can probably be hosted by the 1 meg board. In a longer run a full
> System 3 can be ported to an FPGA. I wonder if 4 megs are enough. I
> am contemplating designing a board with Cellular RAM, which is a good
> compromise between memory capacity, performance, price, and firmware
> complexity. An FPGA based board with 32 megabyte of Cellular RAM is
> quite feasible. Such a board can certainly run System-3.
> 
> W.
> 
> > I have read with more than passing interest regarding comments on
> implementing Oberon 2 for RISC5 targets.
> 
> --
> Oberon at lists.inf.ethz.ch mailing list for ETH Oberon and related
> systems https://lists.inf.ethz.ch/mailman/listinfo/oberon