Older blog entries for connolly (starting at number 25)

I'm no longer happy with my personal wiki; since upgrading to a version of zwiki that supports dated comments, I find that I'm more comfortable doing "episodic publishing" (i.e. blogging) than collaborating on collected wisdom. A personal wiki is an oxymoron; the wiki genre is all about collaboration.

So I started thinking about how to migrate my content to a blogging system. The first step was to somehow grok all the data I've got in there. I asked in the #zope channel about .zexp format and was discouraged from peeking inside. I was advised to write an external method that runs inside Zope to export my data, but by the time I saw that advice, I already had dumpwiki.xsl converting the zope XML export format to XHTML. The actual contents of the wiki pages was quoted, so I'm undoing that with python and xmltramp. I think I hit a bug in the xmltramp serializer. Gotta look into that.

Anyway... the #zope guys were surprised that I had never done any Zope methods. I explained that I use Zope because it's the only server you can apt-get and write to (with iPhoto or emacs eldav mode) out of the box, with elephant-never-forgets versioning.

That got me thinking... with the "cvs is good enough" orthodoxy eroding, and all the work on subversion and arch and darcs, maybe it's time to take another look at the versioning part of WebDAV. Especially git, with its cryptographically secure history... because the problem with the writeable web is more social than technical. People use sftp rather than HTTP because the social protocol is well known, not because FTP is a better protocol than HTTP for writing.

Hmm... if I had a PAW blog or a DIG blog, this might belong there. Maybe reltag will work...

I sure wish python had more penetration in the blogging world. PHP is clearly the server-side deployment vehicle these days, and javascript is the way to juice up clients, but neither PHP nor javascript meets the unambiguity requirement that I think is critical for software engineering in the large.

Got tired of the manual login-and-download-statement ritual with my bank, and since ClientForm and ClientCookie are such a joy and python has SSL built-in, I cooked up grabst.py that automates it. Heaven forbid the bank website should allow me to just bookmark my statement so I could directly GET (with SSL and password auth) it, without all the frames and javascript malarky.

Glom reports looks interesting... using XSLT and CSS for reports... GTK... python API... postgress; that reminds me that In quacken I have code to transfer all my Quicken data (about 10 years worth) to postgress in one go for use with an old version of saCASH. saCASH development has since gone in a direction that I'm not so interested in, but this glom thing looks like a good match for my FractalAccounting goals.
Grokking Triples from Spreadsheets

Sean notes that there are lots of triples in spreadsheets. Yup. After my Aug 2003 trip to Montreal for Extreme, I used gnumeric as an RDF authoring tool to collect all the gas receipts and such; then the Makefile has this stanza to convert it to RDF:

triplog.rdf: triplog.xml grokSheet.xsl
	$(XSLTPROC) --novalid grokSheet.xsl triplog.xml >$@

I haven't scrubbed the data, so this is somewhat incomplete as a demo.

Yes, this is another GRDDL style transformation.

A comment on Sean's blog said "don't forget RDBs". Of course not. See Relational Databases and the Semantic Web; I hope to update my implementation, dbview.py to use SPARQL before too much longer.

Hmm... where are timbl's slides on RDF, trees, tables, and such?

WordPress Struggles

Famous 5-Minute Install "2. Create a database for WordPress on your web server, as well as a MySQL username to access it." Umm... that step alone is taking me a lot longer than five minutes. The debian package doesn't seem to get along with Dom's installation:

Installing new version of config file /etc/wordpress/index.php ...
ERROR 1007 at line 1: Can't create database 'wordpress'. Database exists

IntegrityIsJobOne does gnome blog know this? they advertise... "Entries can be written gradually over the course of a day, popping gnome blog open and closed as you have thoughts to jot down and then posting at the end of the day"

What if my computer crashes in the middle of the day?

19 Feb 2005 (updated 19 Feb 2005 at 03:31 UTC) »
testing gnome blog

My son is supposed to keep a reading response journal. I wonder about having him do it with a blog... maybe using livejournal's friends-only authentication stuff... or better: is there an easy-to-administer blog thingy for zope?

I keep bugging TimBL and EricM about blogging (esp with trackback and comments) too.

This posting got kinda mangled; the title got thrown away.

Hmm... hypertext UI? my home page hmm... klunky dialog interface, but the resulting text is OK.

Suppose I drag n drop from my web browser? Ugh. lose. Where's paste link when we need it? Hmm... recoverable... gnome blog

19 Feb 2005 (updated 8 Mar 2005 at 18:51 UTC) »
taking the plunge to Ubuntu Hoary

For those of you who have missed the daily churn of new packages, Hoary is for you.

-- GuideToHoary

They've got me pegged. I've been scratching the itch with some 3rd party repositories and having trouble. If Hoary is just two months from release, I'm sure it'll work well enough for the kids' PC.

OS X keychain trustworthy?

Every time I log in with SSH Agent, that little "add to keychain" checkbox is staring at me, but I don't know if keychain source is open to the security community, and I don't understand its architecture. Any reading recommendations? Maybe drop them in the del.icio.us keychain tag.

postscript: yes, it's trustworthy. under the hood of the OS X keychain -- 19 Feb 2005. Love having a separate keychain for financial stuff. Camino groks, but Firefox doesn't. But Camino doesn't grok type-ahead-find... BrowserChecklist

Building a PC... end of an era?

The kids' PC was noisy and slow. The Mac mini buzz was very tempting, but after going over Micro Center specials and a linux hardware buying guide with Brennan, we chose to build/upgrade our own, for under $150.
-- Building a PC

I wanted to pass on this experience to my son before Hollywood makes it illegal (due to DCMA++, Trustworthy Computing, ...) or something else destroys the commodity PC hardware market.

When I eventually got the thing working, the boot time is long enough and it's quiet enough that I'm thinking of leaving it on all the time. And it's orders of magnitude faster than the machine it replaces, making me wonder what to do with all this compute power, beside cartoon network games and tuxracer...

Family Finances and Entertainment, Ubuntu, and Debian

I think the machine should make a nice PVR (or at least to convert VHS tapes to digital video); too bad there are no Ubunutu packages for freevo nor mythtv. There are external debian repositories with freevo packages, so I went to install debian on the 2nd disk, using a netinst CD I burned just a few months ago. It tried to get a DHCP lease using the ehternet controller on the motherboard, but the connection to this machine is wireless, thru a Microcenter: D-Link DWL-G520 that I got for $27.99 after rebate. It's listed among the madwifi supported hardware; of course, the madwifi HAL isn't open source; Ubuntu supports it via a restricted-modules package, but the debian installer does not. I considered building a kernel module from source, but I have a bad taste in my mouth maintaining a WinFourLinOnDebian kernel. You see, I still rely on Quicken for family finances; the open source alternatives are missing features that I rely on (budgeting, quick-zoom reports, ...).

I once dispaired that open source would ever produce a competitive web browser, and that day has clearly arrived, so I remain hopeful that open source will conquer the wifi and PVR applications, though the challenges are formidable.

End of year giving

Hmm... which to give to... EFF? SPI? United Way?

Why did eff.org take forever to respond? spi-inc.org is refusing connections. debian donations prolly works.

16 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!