dev-libs/memcache-api-php is gone

When this package was added it was intended as a stopgap measure, really. There was no good PHP API for the memcached daemon. The only good one was the Perl one, (now Cache::Memcache) written by the memcached author.

There was later a PHP API written in C, part of the PECL, which would inherently be faster. It’s now the recommended API (SO USE IT, AND DISTCC! </subtle).

So now the slow, buggy, unsupported API is gone. Rejoice!

The UO Addiction has Gentoo implications

Cos I’m about two seconds away from comitting UOAMhub, an unofficial Linux port of the Win32 only UOAMsvr.

UOAM is a program designed to act as a kind of GPS within the UO world. Various guilds and groups of friends use UOAM to keep track of everyone’s position, which, coupled with voice communication (such as TeamSpeak) helps in coordination of events.

Belxan’s UOAM protocol is proprietary and the uoamhub daemon has reverse engineered it.

I’ll be testing the daemon today and stress testing it this week with my own guild.

sys-libs/memcached-api-php removal!

I posted on the -dev mail list late last month about removing this API. It’s buggy, and the source files aren’t even on SRC_URI. There was 0 response to my -dev post and 0 favourable response to my post to memcached mail list.

So if you’re going to use Memcached with PHP please use the PECL version. dev-php5/pecl-memcache or something. 😉

Things I’ve learned by using Gentoo…

In my years using Gentoo I’ve learned a lot. I’m going to share it with everyone (whether you like it or not):

In no particular order…

  1. It isn’t necessary to completely wipe clean your disk to fix a perceived problem in your installation (even if you allocate only 4gb instead of 40 for /).
  2. Never downgrade glibc.
  3. Always use Google when you’ve got a problem.
  4. ALWAYS use Bugzilla
  5. Join Gentoo🙂
  6. Ignore flamewars
  7. Don’t ignore -core mail or you’ll miss out on important data, like, Gentoo is a foundation? And it’s got a council? *g*
  8. Disable root logins, ssh version 1, and password logins in sshd_config; force ssh key access
  9. Backup your data
  10. ADMIT NOTHING!!!!!!!
  11. Don’t take things too seriously
  12. Take the serious things seriously

There’s so much more, but it would take far too long to list.

More on Distcc and SLP

As much as I’d love to see SLP integrated with Distcc I’ve come to the realisation that Gentoo isn’t the place for the patch to be implmented and tested. Rather, it should be applied to the source tree UPSTREAM for the entire user base to test, likewise with all other patch enhancements.

Obiously it’s different if the package has no maintainer and hasn’t seen the loving hand of a programmer in years or if the patch is to fix a serious Gentoo-specific bug, like the TMPDIR issues in distcc from a year or two ago.

Distcc and SLP

With Bug 80219 we’re going to be adding SLP support to distcc. It will be a boon for those people with large clusters of volunteers.

Right now it seems to work. The daemon starts (./distccd -p 3666 –listen 192.168.0.254 -a 192.168.0.0/24 -P /home/lisa/projects/distcc/distcc-2.18.3/src/distccd.pid –verbose –log-level=debug –log-file=./log.log –daemon) and the tool works (./distccenv —> DISTCC_HOSTS=”lithium:3666 “).

I wonder if the portage guys could do a test to see if slpd is running and use that to set DISTCC_HOSTS with the distccenv program: export `distccenv`. I haven’t done any compilation tests just yet.


On Distcc and bootstrap:

One of the major problems with using distcc during installation from stage 1 is that the user must manually add the distcc user before issuing the emerge distcc command. When the user later (after stage 2) installs distcc the shadow package is available for useradd.

I’d love to just make it so the user has to re-emerge distcc after bootstrap. I think I may have found a MAJOR brain fart: Currently the distcc use is required during emerge to change the permissions of /var/run/distccd to that user so the init script can put the pid file there, owned by distcc:daemon. Inside the init script the directory is chmodded again for the distcc user. Why add do it in the ebuild when it can be done in the init script? *duh* I’ll test that with the SLP stuff.