A variety of things. Today I had the opportunity to read two poems to Bety. One is by Sara Teasdale, probably my favorite poet.
Strephon Kissed me in the Spring,
Robin in the fall
But Colin only looked at me
And never kissed at all.
Strephon’s kiss was lost in jest,
Robin’s lost in play,
But the kiss in Colin’s eyes
Haunts me night and day.
— Sara Teasdale
…and a fun one:
When I am sad and weary,
When I think all hope has gone,
When I walk along High Holborn
I think of you with nothing on.
— Adrian Mitchell
The story of Sara Teasdale is pretty sad. I actually own a very old copy of Strange Victory … it’s emotionally very tough to read sometimes, although I haven’t actually more than cracked the covers of my copy. Haven’t had a lot of time to read poetry lately.
At work, I’ve been scheming on how to get an automation tool that’s simple enough for non-programmers, maintainable, and at the same time, useful. Recommendations welcome. So far experiments with a JS-based recorder of browser-actions went well. It lets me record actions at a browser level, rather than protocol level. All hail Greasemonkey, which is going to encourage me to learn mozilla extensions. So far my recorder is dumping to a wiki for centralized data store, and I put together some auto-script-writing-wizard stuff that uses the uber-l33t XmlHttp stuff to fetch data.
Currently I use XmlHttp to fetch a list of pages at startup, and then per each action, fetch the corresponding page (action) in question and grep it for parameter tags. Those parameter tags (actually python
def funcName( p1="val1" ):) are then presented to the user via popups, etc. as an assistant (like VisualStudio auto-complete?) in generating more complex scripts. Probably will have to move away from JS
prompt() calls and towards a floating <div> that toggles and is dynamically populated with elements, values, etc. (this is especially for the case of multiple parameters on a single action).
Pretty cool being able to conditionally fetch information, parse it either server-side or client-side and do fancy JS / DOM manipulations to make it do what you want to do. I forsee a bright future in this type of work, just so long as it’s testable (should be, since you can treat your XmlHttp requests as requests/responses to be handled appropriately.
Which brings me to my third point of miscellany: XML.
foreach loops is easy enough in any language except Java). The closest thing I’ve seen to native support is the perl lib EasyXML (I guess it got converted to XML::Simple nowadays), where once you’ve loaded the XML object, it gets treated as a first-class perl citizen. Yay. But for now, the whole point is that most developers want access to the first-class citizen, not XPath, and that means that JSON is the path of least-resistance for getting information out to a browser (no dependent libraries). Hopefully it will die a quiet death as more languages get native XML support, but for now it seems pretty useful.
And a final closing funny: “Better to have a nickname than an adjective.” (said by “Big” Dave to “Magpie” ;^)