Notice: Function _load_textdomain_just_in_time was called incorrectly. Translation loading for the math-captcha domain was triggered too early. This is usually an indicator for some code in the plugin or theme running too early. Translations should be loaded at the init action or later. Please see Debugging in WordPress for more information. (This message was added in version 6.7.0.) in /var/www/news.numlock.ch/public_html/wp-includes/functions.php on line 6114
IT – Page 26 – A changelog by Daniel Mettler

watchdog

i’ve set up a “watchdog” to watch http://people.redhat.com/mingo/ using a cron script i’ve written a while ago. it checks ingo’s patch site for changes once per day and notifies me by email about updated patches. like this, i can better follow the changes and keep my kernel current (always using ingo’s latest patches).

actually i could easily further automate the thing. it wouldn’t be difficult to write a script that checks if there are any new patches, if so, automatically downloads them, reverts similar earlier patches (in the right order), applies the new patches (in the right order), configures and compiles the kernel, if successful installs it and reboots the box. hmm.. perhaps i’ll do this later. would need to take special care about potential security issues (ingo doesn’t sign his patches yet).

the current watchdog script would also be suitable to minutely watch and track even minor changes on e.g. blogs (remember recent disputes about dave winer’s blog being tracked closely). of course i’m not big brother, so i won’t do this anyway (i wouldn’t want my blog to be watched that precisely myself as i often publish temp/unfinished entries i might want to remove, correct or finish later).

regarding the script: yes, push would be better than pull, but automated pull is at least better than manual pull. currently, it’s basically a faked push service (change/time-triggered e-mail msg) which could be made almost as accurate as a real push service by decreasing the interval at which the sites are watched. effectively transforming pull to push.. and generating page hits like crazy.. gee, darn numlock.ch-effect ddos’ing ;)

/dev/misc

* a premiere: this early morning, i’ve made a linux kernel patch for the first time (not counting mixing/editing others’ kernel patches etc. ;) as ingo asked me to fix the minor devfs issues with his exec-shield patch myself. admittely it’s a very tiny one and i probably couldn’t have done it if ingo didn’t tell me it was a matter of changed field names only. nevertheless it was a very cool experience. dealing with the kernel is just magic :) the funky thing is that i’ve done this all on the box which has just served you this page – i bet nobody noticed anything of me compiling kernels in the background :) currently i’m running 2.4.22-ac1 with exec-shield-2.4.22-ac1-nptl-D4. i will merge in new patches as available.. need to get some more kernel books and stuff.

* i’ve just learned that toshiba is selling now an all-in-one wlan hotspot box built on gentoo. brave move. i wonder how they circumvent gentoo’s rare but still existing and sometimes very annoying quality assurance troubles (critical bugs, blockers and such). a year ago, i’ve been thinking about an all-in-one-barebone with pre-installed gentoo too, but for soho use only (file repository, application server, mail server, dbms, router, fw etc.). later i’ve bought my current lex barebone to do some prototyping. all in all i think gentoo would be suitable for such a thing, if there just weren’t these darn qa issues gentoo struggles with. regarding the hw, i probably wouldn’t use a lex barebone. as an advantage it has a small price tag, but the two fatal hard disk failures (two different brands) i’ve experienced so far (within about 9 months) make me conclude that the lex engineers probably didn’t test their product thoroughly (both hdd died of overheating). for me it’s not that much of a problem (living at the bleeding edge i’m almost used to failures), but for shipping it to customers it’s way too risky and costly.

it’s like.. gone

my main workstation’s “soundblaster live! 5.1 digital” sound card seems to have stopped working a couple of days ago (don’t know why). the “funny” thing about it: whereas linux simply “disables” the sound device, windows (xp prof) “disables” the device but nevertheless freezes after about 5 minutes from boot-up (that’s a reproducible behaviour). there isn’t even a bsod, just “ice age” forever. my wild guess: could be an uncaught kernel counter overflow or sth similar. however, it might just as well be an unrelated problem of course (though i hope it is related in fact). hardware failures are keeping me pretty busy lately.. let’s see what happens if i remove/replace the sound card..

[update: after removing the sound card, windows stability probs disappeared. now i’ve taken a close look at the card and as i couldn’t find any evidence of a short-circuit (braised conducting paths are pretty easy to detect), i’ve reinstalled it. believe it or not: *drum-roll* the card is recognized again and all the woes have gone :) might just have been a loose connection..]

various bites

tried to apply ingo molnar’s exec shield patch to 2.4.22 vanilla sources but there were 3 (minor) rejects. i hope ingo is still maintaining his patch as i don’t want to miss it anymore (“put an end to your nightmares, get general buffer overflow protection for your server” ;).

reminds me that i actually should

* update some apps on this box
* eventually migrate all the old log entries and try to regain those i lost due to the hdd breakdown

the second task needs to wait, i guess ;)

meanwhile, .GNU has started teething. i wonder how much ms paid for system.windows.forms ;) most of you probably agree that regarding development costs, oss will always win those battles (note the pun, heh ;)

[update 20030828@02:30am: ingo has just sent me a recent version of his exec shield patch (exec-shield-2.4.21-cambridge-D2) which according to him should apply to most 2.4.22 kernels. of course i’ve instantly tried it with a 2.4.22 vanilla kernel – unfortunately there are 11 rejects. hearing he will do a 2.4.22 port soon is excellent news however.. many thanks to ingo for his efforts and this kind and obliging response!]

impolite feedster

unlike google, feedster doesn’t care about robots.txt policy files at all (i’ve forbidden indexing of /news, yet it did). netiquette doesn’t seem to count anymore.

the question that arises is whether this is

a) the result of missing (oldschool) experts in feedster’s dev team(?) or
b) intention

i am close to stop blogging under such conditions. notification of central changelogs such as blo.gs is ok (it happens knowingly and actively), but i don’t want my random thoughts be spilled all over the net.

What’s so bad about “the most sincere form of flattery”?

I wonder how long it takes until “web shortcuts”, a genuine [Update: probably not genuine, see the comments below] feature of Konqueror [1], are imitated by any of the other browsers (i.e. Mozilla Firebird, IE, Opera etc.).

Here’s how Konqueror’s web shortcuts work:

To search the web using Google, simply type “gg:” followed by any Google search term(s) in Konqueror’s location bar (that’s where URLs are entered).
To look up a term in Dictionary.com, it’s “dic:” followed by the term. For Freshmeat it’s “fm:”, for Foldoc it’s “fd:” or “foldoc:”, for IETF RFCs it’s “rfc:”, for OpenPGP keys its “pgp:” etc. you get it ;) (BTW: You can even search MSDN using the obvious web shortcut! ;)

Predefined web shortcuts can be edited/deleted, customized shortcuts can be added/defined (among others I’ve added some shortcuts for Leo, Linguadict, Gentoo’s Bugzilla)

For my daily use, Konqueror’s web shortcuts concept is superior (more convenient and faster) to any other web shortcut concept I know of (e.g. Bookmarklets [IE, Mozilla], Sidebar [Mozilla], Hotlist [Opera], having separate entry fields for each shortcut [Galeon]. The latter being particularly.. err.. “special” IYKWIM ;). If I’ve missed a concept, please notify me. [Update: Mozilla has a similar (hidden) feature too! Thx to Asa for the hint]

You might call browsers a “non-issue”. I don’t. “Web shortcuts” is only one feature picked out of many other features that still differ between browsers. Some of these differences are details only, but they can make a big difference when it comes to usability.

So, why hesitate imitating as long as it’s legal?

[1] Note that web shortcuts have been part of Konqueror since at least 2001-08-15! (Sorry, can’t find the first check-in of “web shortcuts” code in CVS)

‘gotmail’ slightly updated

btw. i’ve slightly updated the ‘gotmail’ script. now it features coloured output which makes it better suitable for console usage. if you pipe its output to a text-to-speech processor you might need to disable the color codes. the most recent version of the script can be found here:

https://news.numlock.ch/docs/gotmail

as an addendum to “Make your server play music and talk..” you should modify the crontab entry the following way to prevent cron from sending mail messages (this can be achieved by piping any stdout/stderr output to /dev/null):

e.g.
/etc/crontab:

# mettlerd: make hal read aloud my new mail every 10 minutes
*/10 * * * * mettlerd /usr/local/bin/gotmail|festival --tts >/dev/null 2>&1

[Update 20031101: minor fix to make gotmail display mailinglist digests correctly: use egrep’s “-m 1” option to prevent multiple listings of from and subject lines]

b2++ :)

rather coincidentally (*) i’ve noticed that donncha has commented on my recent question about b2, recommending b2++ instead of b2. well, b2++ certainly looks promising :) so if i will evaluate alternative blogger soft (it’s a matter of time only) i will probably evaluate at least b2++ and wordpress (the “official” successor of b2). both will be superior to mt for sure.. e.g. mt’s need for “rebuilding” the static pages after every tiny bit of site editing is pretty annoying imho.

(*) actually i’ve enabled mail-notification in mt and my smtp setup just works perfectly. nevertheless i haven’t received any notification messages so far. i suppose something must be wrong with either mt itself or my mt setup..

Fanless II

Now my graphics card (an old Geforce 2 GTS) is fanless too ;) Basically, I just removed its annoyingly noisy fan, leaving the passive cooler block only. I wonder how long it will withstand the heat :> Uhm.. actually this shouldn’t be a problem as there are similar, even higher clocked chips which are sold fanless today.. (AFAIK my Geforce is clocked at 250 MHz)

Finally, my main workstation is “silent” :)