Older blog entries for sab39 (starting at number 18)

caolan: That's cool, but I disagree that it can't be done smaller...

sballard@rainbow:~$ ./test; echo $?
42
sballard@rainbow:~$ wc -c ./test
     18 ./test

Of course, I cheated...

sballard@rainbow:~$ cat ./test
#!/bin/sh
exit 42

Badger: Someone on one of the Moz newsgroups posted a fast-update perl script - it would use the Bonsai webserver to identify the files that had changed, and then use CVS to download those only, bypassing the usual CVS negotiation of these issues. The discussion suggested that this really helped, although it seems to me that CVS really ought to be more efficient at this kind of negotiation if a *webserver* can provide the required information so much quicker. Try looking through n.p.m.builds (and ask if you don't find it).

lkcl: See EVE - a massively multiplayer space/strategy 3D game with the servers written in Stackless Python. Incredible screenshots, etc. No mention of whether it runs on Linux that I could find, which seems strange when they're using an open source language. Now if only { I had a decent graphics card || Xfree supported acceleration on the crap one I do have } ...

Election day! Of course, I can't actually vote, being only a permanent resident for the last 2 years and at least another few years away from even being eligible to consider thinking about applying for citizenship. The phrase "no taxation without representation" has a really delicious irony for me...

If anyone who wasn't planning on voting feels like rectifying this heinous situation by giving me their vote, please cast it for Nader :) And remember, this only helps Bush if you were going to vote for Gore otherwise...

Does anyone here know of any site with Nader's exit poll results? I'm not terribly interested in the Bore/Gush race (and I can see why releasing those results early is potentially unwise... so thanks to whoever is DoS'ing drudgereport.com) but being able to track how close Nader is to pulling in the 5% he needs would be great...

I'm reading "Stranger in a Strange Land" for the first time. Absolutely great book, and I'm amazed I've never read it before. The whole "life on Mars" premise is dated, of course, but it doesn't even matter.

The only reason I'm mentioning this on Advogato is that it's really weird to read the word "grok" and think that the reader wasn't supposed to understand what it meant right away. I'm just getting to the point in the book where I'm beginning to realize that I don't understand a fraction of what it means anyway, despite having been using it for a year or two...

Whatever you do, do not run the following command:

mv /lib /test-lib

Example...

# mv /lib /test-lib
# mv /test-lib /lib
/bin/mv: No such file or directory
# ln -s /test-lib /lib
/bin/ln: No such file or directory
# perl -e 'rename "/test-lib", "/lib"'
/usr/bin/perl: No such file or directory
# reboot
/sbin/reboot: No such file or directory

I think you can see where this is going...

Of course, on reboot you can't even get as far as running init, even if you set init=/bin/sh. I hope you have a good boot floppy.

My 2c on the KDE business: I've defended Debian in the past (in particular arguing the finer points of the licenses involved with ralsina) for not including KDE, but I will say this: If debian do not include KDE now (or in the near future), then I hereby retract all defending I have ever done of their position. Fortunately for my own faith in humanity, I see no indication thus far that they won't. Whether kmidi and kghostview get incorporated without rms's "forgiveness", I'm not sure, but I really don't see any of the rest being omitted.

As far as kdelibs still being in incoming after months and months - how many people actually want to install kdelibs without the rest of kde? It would have been nothing more than a symbolic gesture, and as such, I'm only a little disappointed that nobody decided to go ahead and make it anyway.

What would really do wonders for my faith in humanity would be for the authors of ghostview and timidity to come out of the woodwork now and grant "forgiveness" - regardless of whether it's needed and/or whether the KDE people want or accept it - just so that the flamewar can be over. It's interesting to me that there's been no comment from either group thus far...

I don't know whether I think this "forgiveness" is really needed or not - the GPL does seem to imply it, but AFAIK no previous GPL violator has ever been taken to task over it, which in itself might make that clause unenforcible. And at this point - does anyone really want anything other than for this to be over?

Raph wrote: 1. Not checking error codes. Virtually all invocations of functions or commands should be of the form of: status = invocation(...); if (status) { cleanup() }

UGH! In some ways I agree with you about code reuse, but after reading comments like that it makes me remember why I never use C anymore, in favor of Java. I could live without almost all of its features, but I can't imagine writing code without exceptions anymore. You can still be lazy and write bad code - just catch and discard all exceptions at the top level - but you have to work at it. And instead of having every useful line of code be followed by 3 lines of checking status, you can put it all in the exception handler.

I wish that it were possible to do real work in Java without resorting to non-free software, though... For my day job, I'll settle, but not for hacking. I look forward to the day when Kaffe, Classpath or libgcj are really up to the level of Sun's implementations and I can ditch that proprietary bulls**t. Unless Sun follow their StarOffice announcement with another one...

I don't think the other aspects of Java are completely useless, but I do agree that they're no kind of silver bullet. I find that (implementation) inheritance is most useful within a single project - making inheritability a part of your public interface is much harder than it looks. Even Sun's collections API, with its Abstract* objects which are designed to be inherited from, has problems here... you need to know altogether too much about the implementation of each method to know exactly what functionality you must override for your specific implementation. For example, if you override add(Object), will addAll(Collection) know to use your version, or is addAll implemented its own way? What about delete(Object) versus clear()? The issue is compounded by the way they handle threadsafety, which is far too complex to explain here, but isn't documented nearly well enough to understand exactly what you need to do if, for example, you can provide an optimized addAll().

Generics (real generics, not bloat-inducing templates) would really help java too, and a complete rethink from scratch of all of the core APIs. But the language itself is pretty nice for avoiding the more basic kind of quality pitfalls, and some of the APIs are too.

jeffs: Have you looked at GNU Classpath, Kaffe or gcj (part of gcc, but I don't know it's URL)? Those are the primary free software java projects that I know of, and I'm sure any one of them could use some help :)

nullity: If other hackers are like me, the reason they haven't read your first chapter is because, if you only read the recent-log, all you know about it is that it's "the first chapter of a book", not what it's actually about. And if they're like me, they're too lazy to click the link and find out if the subject is one that they're interested in :)

rms got back to me about my license with major issues :( Mainly his point boils down to the fact that legalese isn't easily "patchable", so writing it as "The GPL, but with the following exceptions" didn't work. I'm now trying to write it from scratch, taking the relevant parts from the GPL.

Disclaimer for all those about to flame me: Yes I know that "yet another license" is a bad thing, but I'm writing it to be explicitly compatible with every other Free Software license. As far as I know the only license that acheives that right now is the modified BSD license, which is also compatible with all proprietary licenses... If anyone can point me to a copyleft license that still provides compatibility with everything from the MPL to the QPL, I'll be very grateful to stop trying to write this legal mumbo jumbo :)

The whole "unix sucks" issue seems to me to be a battle to state the obvious in the most provocative way. Has nobody ever read Worse is better? (the relevant section is in the middle somewhere). Unix does well because it sucks, not despite it. This is also why Linux is outstripping the *BSDs and why Windows beat Mac. The author clearly had an incredible degree of foresight - the comment that "Fortunately, in 1995 we will have a good operating system and language. Unfortunately, they will be Unix and C++" is incredible, considering that in 2000 what we have is Linux and Java.

Speaking of Java, did anyone else find this comment telling: "If GPL is good enough for StarOffice software, then why not for the Java[tm] platform [...]?" That is a good and valid question, but let's not lose sight of the main point: First and foremost, GPL is the right thing to do for the StarOffice suite [...]. If you're sufficiently optimistic, it almost sounds like they're saying "That's a good question... wait and see". Maybe I just really want it to mean that...