Paul's Journal

So, my new workstation at work decided today was a good day to stop working. While I had my full IDE open, it detected journaling errors, and decided to re-mount everything read-only (say goodbye to changes). Rebooted it, now watching in pain as fsck tells me my filesystem sucks:

Inode 2003227, has a bad extended attribute block 5276170. Clear? yes

Error reading block 4097088 (Attempt to read block from filsystem resulted in short read) while doing inode scan. Ignore error? yes

Force rewrite? yes

[continue, forever.]

Running FSCK always sucks. I wish I could use XFS, but RHEL4 only includes ext3. Too much pain today. Too much.

At BarCamp:

“Tagging is cool, but I haven’t found a use for it.” -Nick

I am over at BarCamp in Palo Alto right now, prolly hang out here Saturday and Sunday. Anyone else coming?

WordPress recently released 1.5.2. The problem is that they modified the release tarballs, after putting the announcement on their site.

Don’t do that. Once you make a version public NEVER change the release tarballs. It doesn’t help that they didn’t sign their releases or even provide basic hashes, but the fact that now some users who installed 1.5 2 are still vulnerable, while some others are not is the problem. This defeats the one major purpose of versioning — to help the users and admins who install your product. So they can know if they are vulnerable to an issue.

I am glad that I don’t use WordPress, this kind of behavior is not excusable for an Open Source Project. Its about providing trust to your users. I can’t trust that a WordPress 1.5.2 install is secure, since this one version has two different contents.

Dear Brad,

memcached is awesome. I owe you at least a few beers.

Thanks,

Paul