evan_tech -- livejournal ruby module update

04:15 pm, 5 Mar 06

livejournal ruby module update

My LiveJournal Ruby module now syncs comments. I had this mostly working for a while but finished while awake at 5am, jetlagged.

You now need libxml-parser-ruby because REXML is hella slow (like 10 seconds to parse the 400k XML file that LJ sends you -- I must be doing something wrong).

Unfortunately, I forgot to monotone sync my computer before I packed it up and so I don't have my recent changes to Ghosd. Do I let it wait six months (when I get my computer back) or do I just work from what I have?

teferi	You're not doing anything wrong, REXML is hella slow. Great API, though.
reply to this

evan	Yeah. It's so pleasant! (Well, as pleasant as XML can be, anyway.)
reply to this

teferi

This is one of the things about XML that bothers me. Everyone says it's great because it's easy to machine-parse and -create, AND easily human-readable and editable.

But:

- it's a pain in the neck to read and edit by hand (ever twiddled fontconfig options? Ugh.)
- most of the XML parser APIs (and creation APIs?) are painful.

So...what exactly have we gained by embracing the holy angle brackets?

reply to this

evan

Nothing. I think, generally, that people who know better just avoid it (like they do other popular-but-not-ultimately-useful technologies like PHP or RDF). For example, I don't think we use XML within Google for anything except where we're forced to (interacting with RSS, accepting feeds).

I think for most tasks people use XML for, YAML would be more pleasant. But it unfortunately has a gigantically bloated spec...

reply to this

dan_erat	There's at least one (big) database schema that's described in XML. As for waiting six months to work on Ghosd... yeah, right. I can't even manage to go to sleep when I'm working on something fun. :)
reply to this

edge_walker

I disagree quite strongly. For document markup, SGML/XML-style markup is actually a lot more pleasant than any other options I’ve encountered. Try imagining an HTML document rendered as YAML. Fun? Not my idea of it, anyway. Or maybe you think LaTeX is more readable? In that case we’ll just have to agree that our tastes disagree wildly. :-)

Really, if you consider the field which bore XML, née SGML, the syntax is not at all surprising.

Another thing is that doing XML means that everyone involved must have met all necessary (though not sufficient) preconditions for internationalization. XML’s the horse on whose back Unicode rode in, to a great extent.

As for APIs, well, let’s say I used to hate XML until I learned XPath. :-) For manipulating documents, I’ve found the XSLT model to be very nice, though you’ll have to hold your nose for a long time until you get used to the verbosity of expressing the syntax in XML. What I’d agree is currently crap is that there’s no concise, humane API for creating XML documents from scratch, except maybe the Kid template engine (but that’s in Python so I have no use for it; cue plans to write something very close to it in Perl…).

That’s not to say XML is a panacea. It has warts, like every format, and people abuse it for all kinds of crap that would be better serialised in a different format. All told, though, I think it’s a pretty damn good choice for the things it’s good for, despite the fact that that list excludes being one-size-fits-all.

reply to this

heydusty

There has been some work done to optimize forthcoming versions of rexml* but it's mostly been memory consumption which has been more of a problem for me than sheer speed. It is still ridiculously slow. Slow and beautiful.

Last I checked (about a month ago) I still couldn't build the ruby libxml bindings on my mac. I really wanted/needed them, too, so I ended up working on the project on a server. I think there has been increased activity in the maintenance of the bindings lately (including the emergence of a better set of docs), so I have hope...

* http://www.germane-software.com/projects/rexml/wiki/MemoryUsage

reply to this

evan That's a shame that they don't build!

If you're in the mood for hacking, I'd take a patch to make it use Ruby's libxml bindings, too...

reply to this

evan	That's a shame that they don't build! If you're in the mood for hacking, I'd take a patch to make it use Ruby's libxml bindings, too...
reply to this