[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
> I did it over the summer, so I don't remember what version of
> CLUE I was using. I was using a Beta version of Lucid 3.0 and the
> latest versions of PCL.
Undoubtedly, this was CLUE 1.13. The latest CLUE is 1.15, shipped in September
with X11 R3. Nota bene: CLUE 1.15 *must* be used with R3 CLX.
> The #+explorer things weren't too bad, it
> was when they didn't have #+'s in things, such as references to the tv
> package that I had problems. ... I also had to do
> some obnoxious hacking around with the I/O system in lucid so that it
> would support CLOS instances as I/O streams, and so the reader would
> support weird characters like #\Refresh.
The CLUE spec doesn't make it altogether clear, but the implementation of
clue:interactive-stream is necessarily non-portable. That is, you've got to
integrate it in with your Lisp's stream implementation at a low level. The
interactive-stream code in the CLUE release is good for Explorers (natch) but
only an example for other environments.
This problem would be alleviated if there existed a standard generic function
protocol for implementing Common Lisp streams. This is a worthy task for X3J13.
> All in all, it was the least pleasant summer job of my life and did bad
> things for my self-esteem. I spend so much time fixing CLX, PCL, and
> CLOS that I got virtually no work done on the real project at hand.
This is most regrettable. On the bright side, most of the PCL integration
problems should disappear soon (if not already), as PCL is rapidly improving in
its reliability and its conformance with the CLOS spec. On the gloomy side,
no one has yet committed the energy to picking a good PCL rev and nailing down
the PCL/CLUE integration problem once and for all (where 2 weeks < all <