wikidbase

August 31, 2007

wikidbase is something I noticed on the daily python url a little while back. I was playing with it a bit tonight and I think it might just be the coolest thing in the world.
An instance of that would easily let people input arbitrary data but in a structured way that could then be queried and used. There’s much to lust about there.
– It’s an instantaneous and very flexible issue tracker
– it’s a kick ass todo list manager
– it’s a hw functionality indexer
– it’s a rolodex
– it’s an enemies-list
that’s the cool part – being able to easily input one of the ‘types’ of pages you create and add more and more like it using the format you created the first time. Very hot.

the help page  explains a lot more. – the demo login is admin/admin

Advertisements

 The fedora package update rss feeds have been down for a while. I got around to getting them back up today and now you should have them available here:

http://planet.fedoraproject.org/infofeed/

pybackpack

August 22, 2007

for reasons I don’t know I ended up playing with pybackpack tonight. I had messed with it before but only to test it out – tonight I felt like looking a bit deeper. And it was 2 am – what better time to start on something new, right? So, I posted this note.  The gist of which is – there are somethings that would be great additions to pybackpack. They are:

  • easy addition of files/dirs to a backup set from nautilus
  • gnome applet to nag you about how long its been since your last backup
  • rsync.net, amazon-s3, etc integration
  • using duplicity instead of rdiff-backup to allow for encrypted backups

then I proceeded to look at solving some of  these items. After an hour of researching how to do it (nautilus-send-to? no, nautilus-extensions? closer, nautilus-python? ah, yes, docs, hmmm…, source? ah, better) I put together this little python extension to nautilus.  Now, it is very simple – but it lets you quickly add files or dirs to a pybackpack backup set using nautilus.

If anyone has some interest in hacking on the others ping at me – I’d love to talk with you some about them before they get implemented. Something that could be released in an updated pybackpack or a pybackpack-addon package in fedora would be terribly cool.

it’s just an infatuation

August 19, 2007

It’ll pass. I know it will but oh my god these bikes are gorgeous.
I mean, just stunningly gorgeous.

I mean look at the headlights:

kogswell

yum transaction journal

August 17, 2007

I just got finished checking a useful feature into yum. It’s a transaction journal. I have talked about the idea before but I hadn’t really sat down to do the other infrastructure work to make it happen. This last week I did that and it really wasn’t very difficult. Here’s how it works:

When you setup a transaction yum writes out a complete list of each individual action that is going to occur. So if you say:

yum install foo
then the action is:

install foo-v-r.arch

or
yum remove bar

then the action is:

erase bar-v-r.arch

or

yum update baz

then the action is:

install baz-v-r.arch
remove baz-v-r.arch

That’s right – in rpm every update is actually an install of the new pkg and an erase of the files from the old pkg which don’t overlap.
So yum writes these out to a file before the transaction. Then once the transaction starts it writes out each of the items it has completed.
The idea is that if your computer dies, or X dies or the power goes out or yum aborts or rpm freezes or aliens invade – and any of these things kill your transaction – you can just look at what was proposed, look at what is left undone and do the rest. I’m currently working on some code to add to yum to check for undone transaction elements so it can prompt you to finish them.
Anyway – I hope this feature is completely useless to most of you b/c you never have an aborted transaction. But it seems like a nice thing to have just in case