Older blog entries for dwmw2 (starting at number 108)

kaeru writes:
Assign mailto: for Firefox to launch new mail window for evolution, add this line for Mailer:

evolution-1.5 mailto:%A?Subject=%S&Cc=%C&body=%B

Er, why mangle the mailto: URL you were given? That's just bizarre.

Stick to 'evolution %r' and you get the rest of the headers preserved too. Some mailing list archives give mailto links with In-Reply-To: headers intact, etc. and your version would gratuitously discard them.

louie writes:
no one should say 'RH is doing it badly' without an understanding of the difficulty it entails

If I have an understanding of the difficulty, and I'm still dismayed by the length of time it's taken to do anything concrete, am I allowed to say it then?

louie: It is indeed a hard task. But I wouldn't say that the aim is to create Debian. I'd say that is one of the things we need to avoid.

Not that I'm saying Debian is a bad thing, mind you -- but I certainly hesitate to suggest that the world needs another Debian.

Been playing with Fedora Core 2 on my shiny new toy.

It's mostly working fine apart from the fact that the installer doesn't quite manage to make the thing bootable, and doesn't make an Apple Bootstrap partition if you let it autopartition. Switching to tty2 after the install has finished, when it wants to reboot, and then setting up yaboot manually, should get you round that. There are a few remaining annoyances but nothing too major. I made a tracking bug for it: #121179.

There's a handful of packages for FC2/ppc in a yum repository at ftp://ftp.uk.linux.org/pub/people/dwmw2/fc2-mac. There's some Livna RPMS including Xine fixed to work on PPC, pine, and qemu. There was OpenOffice.org 1.1.1 for a while but now that's in the rawhide tree so I removed my build.

Qemu is impressive -- it runs i386 acroread very nicely, and does it inside a Mozilla window with mozplugger. If only I could get a flash player to work similarly. I do have the standalone flash player app working, but the sound is wrong-endian. I wonder how hard it would be to make a mozilla plugin which uses qemu and thunks in and out of native code, to allow you to use i386 plugins...

hub writes:

In fact we don't bounce [mails to nonexistent recipients], we reject them at SMTP time with a 450 error code.

That's bad too. That's a temporary failure. Which means the mail stays on the remote queue and they retry it again after a while. They do this over and over again until a week or so has expired, at which point they decide the 'temporary' failure is in fact a permanent one.

It also means that people who do callouts to verify the sender of mail they're receiving might have been accepting mail with your addresses faked, rather than rejecting it as they would have done if there'd been a permanent (5xx) rejection from you.

If rejecting mail to unknown users with a temporary failure code is, as you say, "the standard behaviour", then I suggest you go file that as a serious bug with whatever distribution you're using. You effectively brought the DoS upon yourself.

3 May 2004 (updated 3 May 2004 at 17:59 UTC) »

Advogato is playing silly buggers. I made a new post, it seemed to overwrite the previous entry. So I edited it, adding 'WTF happened to my previous entry'. Now I have two posts again, but the former -- the one which had disappeared -- contains the post-edit copy of the first.

Now I'm editing to contain this text. I wonder what will happen.

Must resist... brainwashing...

dammit.

Must try to remember Advogato password. Have been manually extending lifetime of cookie in cookies.txt for the last three years.

hadess: Nat West's online banking system works fine with Mozilla if you fake the user-agent ident string to be something it likes.

Since Mozilla doesn't allow you to do that per-site, you have to do it globally, and then you stop showing up in site statistics anywhere as Mozilla; you show up as whatever you're masquerading as to make Nat West happy.

Apparently the Mozilla people think this is a good thing, which is why they refuse to implement per-site user-agent masquerading.

Since masquerading also breaks the Java plugins, it's probably best just to use Konqueror for accessing the Nat West online banking stuff.

robilad: Don't confuse -ko and -kb. Remember that even if there are false positives occasionally they're trivial to filter out of a diff. CVS won't even complain about them if they change when you commit. Extra information, within reason, is good. Lost information cannot be reclaimed.

I shall decline your kind offer to call you a vegetable right now, but if you ever import my code and eat its CVS tags, then one of your customers comes to me for help with it and I can't tell what versions they've got, I reserve the right to change my mind :)

robilad writes:

RCS tags are evil

I disagree. Why should they give you 'bad diffs' when there's more than one person working on the project? If you're sharing a repository it just works as $DEITY intended, surely?

And if another project imports your sources, presumably they do it with -ko to avoid tag expansion. And if you disagree with that you might just as well assert that linefeeds are evil because some vegetable might work with your source on a DOS box and convert CR to CRLF.... hence you should write all your code on a single line, perhaps?

Finally, CVS tags _do_ give me information which the CVS command can't tell me on demand. When someone sends me some source code, or is trying to describe to me what some vendor has given them, if it has my $Id$ tags in I can refer immediately back to the real version that what they've got was _based_ on, and hence see the _changes_ which were made elsewhere. There's no other way to do it than to embed version information in the file which is going to survive through being merged into various other trees and source control repositories, and make it back to me, via whatever torturous route it takes, intact.

mrcsparker writes:
"Couldn't sleep and now I have ssh-method.c working. Going to clean up the code a bit (still will be a bit hackey when I turn it in) and also add a password dialog box so you won't have to put your password in the uri."

Alternatively, you could just ensure that you detach from your controlling tty to work around the SSH bug, and then SSH will happily invoke ssh-askpass for you if it needs a password/passphrase/skey/etc.

See how Evolution does it when invoking SSH.

99 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!