1) When you move a destination object, symlinks don't follow the target . This leaves "broken" symlinks that refer to nothing. Why doesn't the mv command move these too?
Technically that would involve storing back-references somewhere which would be more maintenance work for the filesystem and/or tools, but the real reason is, a whole lotta software relies on the historical behaviour. Consider the alternatives system used in debian (and adopted by redhat): a symlink in/usr/bin/foo points to/etc/alternatives/foo, which in turn is a symlink to the real binary, maybe/usr/bin/real-foo-1. If you update the second symlink to point at/usr/bin/real-foo-2, you don't want the first symlink thinking "oh, that's been moved/deleted/changed, I'll change too"
3) Symlinks cause all kinds of weirds around chrooted file systems
The problem there is that chroot is a nasty hack, not that the symlink stuff is broken. If you disagree, exactly how would you propose the symlink code could cope better?
I've generally been finding that O'Reilly books need a FAR more critical eye before buying than they used to.
Agreed - I wouldn't judge a book by it's cover, or it's publisher. The grandparent's opinion is interesting, not least because I have been thinking recently 'at least, another publisher who looks to be hitting the same niche as O'Reilly'.
Re:Jonathan Zdziarski's DSPAM claims are bogus too
on
Ending Spam
·
· Score: 1
Surely that means it does scale well? It might be a pain to have a DB backend for a single user, but for scaling up to 1,000s of users, that's exactly what you want.
Actually, the book argues the exact opposite. Spam technology has been forced to change at a quick rate because heuristic rules were derived by human beings. By contrast, the statistical approach (he argues) is more accurate, needs less interference by a human and is impossible to game enough to guarantee profitable spam delivery. He argues that if the majority of email accounts were protected by statistical filtering, whilst spam would still be possible, it wouldn't be profitable.
Nice to see work being made on KPDF. Last year, KDE's document viewing infrastructure was ahead of GNOME's disperate GPDF and gnome-gv. There was the beginnings of a common document front-end with kviewshell, kdvi etc. However I don't know if they've managed to finish unifying it.
Perhaps inspired by this work though, the gnome people have thrown evince together: a UI-focussed front-end for document types.
It's as much of a brand name as it always was. The tablet PC was designed and manufactured by the exact same team as were making X41s before the Lenovo deal.
I'd buy a laptop and run linux on it if I knew there was a model which had a reliable docking station. The IBM laptops look great, but I haven't found a single report where you could hot plug and unplug a running, or hibernated laptop from the station without problems. I don't want to give up my TFT and proper-sized keyboard for extended hacking sessions.
I agree. Giving your variables descriptive names is a good practice
I'll play devil's advocate here. Imagine looking at an algorithm that happens to be incorrect. Variable names could trick you into thinking it works when it doesn't.
It's hard to explain what I mean without concocting an example, but if you abstract away all the intended logic, as captured in comments and variable naming, you are left with the actual logic, which will prove different if the algorithm is incorrect.
If they decided to radically change the debian release process in response to ubuntu, even for the better, sarge would be even more delayed. Better to get it out and work on fixing the process for etch.
I'm concerned that sarge isn't imminent but at the moment I'm going to wait and see (well, do more than that, do my part with bug reports/fixes etc.)
I'm so glad at least one person thinks the same way I do. If someone was to ask me about modern innovations in computer interfaces, the things that would come to my mind would be pane-based navigation, auto placement algorithms, not alpha transparency, noisy backgrounds, 3d spinning window animations, etc.
Slashdot Mirror
Well, sort-of, yes. Solaris 10's zones are in the same ball-park, but it isn't the same take on virtualisation as xen.
1) When you move a destination object, symlinks don't follow the target . This leaves "broken" symlinks that refer to nothing. Why doesn't the mv command move these too?
/usr/bin/foo points to /etc/alternatives/foo, which in turn is a symlink to the real binary, maybe /usr/bin/real-foo-1. If you update the second symlink to point at /usr/bin/real-foo-2, you don't want the first symlink thinking "oh, that's been moved/deleted/changed, I'll change too"
Technically that would involve storing back-references somewhere which would be more maintenance work for the filesystem and/or tools, but the real reason is, a whole lotta software relies on the historical behaviour. Consider the alternatives system used in debian (and adopted by redhat): a symlink in
3) Symlinks cause all kinds of weirds around chrooted file systems
The problem there is that chroot is a nasty hack, not that the symlink stuff is broken. If you disagree, exactly how would you propose the symlink code could cope better?
People forget that Apple's designs are created to be usable first and sexy second.
Not so much forget, as utterly disagree.
I second the Spinellis book. Great stuff.
madness. I have an ibook and a thinkpad, the thinkpad is lighter, quiter, less-warm on the knees, and has a better battery life.
I've generally been finding that O'Reilly books need a FAR more critical eye before buying than they used to.
Agreed - I wouldn't judge a book by it's cover, or it's publisher. The grandparent's opinion is interesting, not least because I have been thinking recently 'at least, another publisher who looks to be hitting the same niche as O'Reilly'.
Surely that means it does scale well? It might be a pain to have a DB backend for a single user, but for scaling up to 1,000s of users, that's exactly what you want.
Actually, the book argues the exact opposite. Spam technology has been forced to change at a quick rate because heuristic rules were derived by human beings. By contrast, the statistical approach (he argues) is more accurate, needs less interference by a human and is impossible to game enough to guarantee profitable spam delivery. He argues that if the majority of email accounts were protected by statistical filtering, whilst spam would still be possible, it wouldn't be profitable.
Nice to see work being made on KPDF. Last year, KDE's document viewing infrastructure was ahead of GNOME's disperate GPDF and gnome-gv. There was the beginnings of a common document front-end with kviewshell, kdvi etc. However I don't know if they've managed to finish unifying it.
Perhaps inspired by this work though, the gnome people have thrown evince together: a UI-focussed front-end for document types.
These laptops are produced by the exact same people as produced the X41 line whilst thinkpad belonged to IBM.
It's as much of a brand name as it always was. The tablet PC was designed and manufactured by the exact same team as were making X41s before the Lenovo deal.
Hmm 50 is an over-estimate (maybe it wasn't when the story was submitted); according to http://bugs.debian.org/release-critical/ there are only 28.
It Is in the UK.
I'd buy a laptop and run linux on it if I knew there was a model which had a reliable docking station. The IBM laptops look great, but I haven't found a single report where you could hot plug and unplug a running, or hibernated laptop from the station without problems. I don't want to give up my TFT and proper-sized keyboard for extended hacking sessions.
sounds like a good plan, although as far as I am aware, the LSB has nothing to do with hardware.
Given that this situation has occurred in the first place, it is clear that Telewest don't have a monitoring policy.
They've had scanners looking at port 25 and performing open relay tests for months if not years.
So it's just sid? The way some of the media have described it as being a totally innovative idea, had me thinking maybe there was a difference...
It can run with a dialog frontend that bears a striking similarity to Ubuntu
Don't you mean debian? I would understand if the d-i team were mightily upset at all the credit for their hard work being attributed to ubuntu.
I agree. Giving your variables descriptive names is a good practice
I'll play devil's advocate here. Imagine looking at an algorithm that happens to be incorrect. Variable names could trick you into thinking it works when it doesn't.
It's hard to explain what I mean without concocting an example, but if you abstract away all the intended logic, as captured in comments and variable naming, you are left with the actual logic, which will prove different if the algorithm is incorrect.
Ubuntu supports roughly 9 fewer architectures than debian.
If they decided to radically change the debian release process in response to ubuntu, even for the better, sarge would be even more delayed. Better to get it out and work on fixing the process for etch.
I'm concerned that sarge isn't imminent but at the moment I'm going to wait and see (well, do more than that, do my part with bug reports/fixes etc.)
Am I missing something? Why wait for google, why not just use a9 to try it?
?
There's no reason to have a seperate i686 branch since there's no benefit over the i386 packages for the majority of programs.
Show me one example where O'Caml beats C++.
ICFP contests:
2004 (ocaml 1st); 2003 (ocaml 1st); 2000 (ocaml 3rd, C++ 5th); >2001 (ocaml tied 3rd place with C);
Others winners: 1998 C derivative, 2002 python (I think C++ actually beat OCaml this year)
I'm so glad at least one person thinks the same way I do. If someone was to ask me about modern innovations in computer interfaces, the things that would come to my mind would be pane-based navigation, auto placement algorithms, not alpha transparency, noisy backgrounds, 3d spinning window animations, etc.