[GTALUG] Google wins over Oracle in Java API copyright suit

Stewart C. Russell scruss at gmail.com
Thu Apr 8 14:46:01 EDT 2021


On 2021-04-07 11:21 a.m., D. Hugh Redelmeier via talk wrote:
> 
> When I was an undergrad at University of Waterloo, we were required to
> use FORTRAN (WatFiv).  I hated it.  I liked the notation of Algol
> better and Algol-W (W for Wirth) was a good implementation for student
> uses.  I even created a bit of a rebellion, but it was put down.

That would have been a hard sell, considering that WatFiv was a
locally-developed project designed specifically for university use. It
produced good-enough code but very quickly, while IBM's compiler would
produce great code but bog down the machine because it was quite slow.

Algol-W seems like it was pretty fast too. I was surprised to find that
there's a (somewhat) maintained Algol-W compiler for Linux:

    https://tiddly-pom.com/~glyn/

It's partly written in ocaml, and is really an Algol-W to C filter. It
seems to have some minor issues building with ocaml > 4.05, but you can
force it to work with

    OCAMLPARAM="safe-string=0,_" make

The few times I need the speed of a compiled language, I'll still reach
for Fortran. C just wants to crash when I'm near it, and having to
remember to start arrays at zero just doesn't work for me.

The only functional languages I've ever used (if openscad doesn't count)
were DSSSL (Scheme with an embedded CSS engine for document processing
in SGML) and XSLT (Scheme [except it's in XML syntax] with an embedded
CSS engine for document processing in XML). Both were utterly dismal and
I only used 'em because I was paid to. Not being able to use loops but
having to write functions that called themselves recursively seemed a
huge amount of faffing about and possibly constituted cruelty to
programmers.

cheers,
 Stewart


More information about the talk mailing list