August is supposed to be the slow time of year. Not! Is there ever a lot of interesting stuff out there. Today we have WS-funnies, OOXML Purdah, Web names, Internet Registry structures, and Ruby metaprogramming craziness.
WS-*, now 6 times better! · I don’t think anything I say could possibly add value to Paul Krill’s OASIS forms six committees to simplify SOA. From Robin Cover, a slightly more official take.
[Enough with that deranged snickering! -Ed.]
Office XML Standardization · There’s so much shouting going on around the effort to get OOXML an ISO fast-track blessing, you might think it’d be hard to find anyone saying anything different. Alex Brown is, in OOXML Purdah.
Naming Things · This is well-known to be hard in the general case; one of the reasons the Web works is that it has a tractable approach to the problem; not perfect, but tractable at the scale of the Internet, which is no small achievement. This doesn’t mean that assigning names to things in your Web apps is easy. In Web resource mapping criteria for frameworks, Bill de hÓra dives deep on the subject.
XML for Internet Registries · There’s a proposal afoot to store some of the basic IANA registries in XML. I’ve been asked my opinion. I suspect that my readers might have something to say, which I would appreciate hearing.
The Methodphitamine · That’s the name of Jay Phillips’ metaprogramming model and his write-up of it.
I’m pretty sure that as a name, it sucks. As an example of the apparently infinite richness lurking under Ruby’s hood, it’s mind-boggling and awe-inspiring.
I’m not sure I actually like it.
Comment feed for ongoing:
From: Tim (but not THE Tim) (Aug 13 2007, at 19:28)
Store them in XML for the simple reason that XSLT can derive the other types of documents if people need them, but it's not as commutative operation - it's much harder to derive good XML from text than it is to derive good text from XML.
[link]
From: Sam McCall (Aug 13 2007, at 20:25)
My problem with methodphitamine (agreed on the name!) is it's not general enough.
You can't do the motivating example: [1,2,3].each { puts it }
[1,2,3].map(&it + 2) works, but [1,2,3].map(2 + &it) fails.
If I have to think about the restrictions of the technique, and how to rewrite my solution so that it doesn't violate any rules, then it's probably not saving any time in most cases.
[link]
From: Ola Bini (Aug 13 2007, at 21:30)
Sam:
I more or less agree with what you're saying, except that your examples are very unnatural. I would say that the &it-form is general enough to work with the way you would normally write things, and in those cases it makes stuff more readable. For example, all versions of the Symbol to_proc trick gets turned into
[1,2,3].map &it.to_s
which I find more readable than &:to_s.
Tim:
When Jay showed me this earlier today, the name was the first thing I commented on. =)
[link]
From: Asbjørn Ulsberg (Aug 14 2007, at 00:37)
I think the XML for Internet Registries idea is good, but I'm not 100% comfortable with how the proposed XML syntax looks like. I guess that's up for discussion, though, although I probably won't invest much time in that discussion. I just want to point out that there's some existing standards that should be looked into before baking a completely new syntax and the new format needs a namespace. Regular XHTML tables might even do in most use-cases, to be frank.
[link]
From: Sam McCall (Aug 14 2007, at 01:09)
Most definitely unnatural :-) Maybe a better example:
words.map &it.downcase
words.map CGI::escape(&it) #doesn't work
It seems confusing to have the two cases written differently. But I could be wrong, I'll have to try it out for longer.
I definitely think it's a nice feature of groovy and worth picking up, and the implementation is brilliant. words.map { it.downcase } as core would be better though.
Is it too late to call it $_ ?
[link]
From: Aristotle Pagaltzis (Aug 14 2007, at 07:41)
Tim, an off-topic request:
You posted several identically titled entries in close succession recently. For someone who is subscribed your comments feed, it can get pretty confusing to follow when you do that. If I may, let me please ask for some distinguishing feature in your titles so that they can be told apart at a glance.
[link]
From: Mike Champion (Aug 14 2007, at 09:06)
On SCA, note that even WS-KoolAid imbibers don't necessarily agree that this new layer of abstraction away from the WS-* formats and protocols is useful, and its relationship to SOA is controversial. See for example http://www.davidchappell.com/blog/2007/05/sca-vs-soa.html
[link]
From: Marcel Weiher (Aug 15 2007, at 09:02)
Higher Order Messaging does away with the 'it' altogether, so you can write things like the following (Smalltalk syntax):
words collect toLower.
but also:
stdout do println:words each.
so the CGI example would also work:
CGI collect escape:words each.
Read the paper at http://www.metaobject.com/papers/Higher_Order_Messaging_OOPSLA_2005.pdf
[link]