Juno Again

Trying my Eclipse Juno again. What caused this adventurousness is that I discovered that my old Eclipse didn't support 'Refresh on Access' (which is when a file has changed on disk, and not in the editor, so you can simply update the editor copy to match the version on disk).

I have the following message of hope: if you switch Project Explorer's presentation from 'Flat' to 'Hierarchical', and you get beach-balled, don't kill it; instead keep your hopes up, go write a blog post, get some coffe, tell Juno to text you when it's done. It can take a few minutes.

Mysterious Chrome Websocket Connect Delay

Again, suddenly your Chrome browser stops connecting a WebSocket, or to be precise, it won't connect it immediately.

These the loveliest of errors, the ones without feedback. There is nothing to go on, no clues to debug. So instead, you irrationally suspect your own code (it hasn't changed). You revert changes to code you did, re-run, un-revert them, re-run, etcetera, swearing.

Luckily for you, this happened before, which helps debugging stuff where there is pretty much no clue what so ever. Takes some time, though, until you feel that deja-vu all over again. So now you wait longer each time before declaring a failure, and sure enough, as you remembered, after a minute or so, the connection succeeds. 

The malady is that you're having a Chrome extension, taking that minute or so to run. I would guess that this is caused by its not being able to connect, or something like that. Maybe a server is down. Maybe the minute-or-so is the time-out. Maybe the connections or failed connection attempts add up to precisely this minute-or-so. Anyhow.

So you disable the extension. (At "chrome://extensions/".) Mysterious delay, no more. That's what you do the first time.

The second time, you delete the JetBrains Chrome Extension. Why did you keep it after the evaluation period expired, anyway? There is even the cutest little animation of the trashcan lid ajarring. Awww.

...I don't know how it suddenly got reenabled. Or if it got reenabled, but worked fine, only to fail later, i.e. just now. Maybe Chrome somehow restored older settings during the frequent crashes I subject it to. 

Quantized Keyboards?

I have before me a Logitech K120, purchased recently because the alt key of the last keyboard gave up and handed in its resignation.

I just seemed to notice something peculiar. And annoying, if true: the keypresses seem to be 'quantized'[1]! Noticed this oddity after something between the spelling center of my brain and the text editor repeatedly failed to spell "Gui". It instead preferred "Giu".

First I suspected that somehow the typing style I use made the I key get depressed before the U key. But after some experimentation, I found the following repeatable behavior: "ui", and "mn", when hit together, for instance by effectively creating a very wide finger out of two of your ten favorite little servants, always produce "iu" and "mn", instead of a mix of both! And there are probably other 'inversion-prone' pairs, but I am just not that interested...

Maybe there is a probe pulse frequency of perhaps 100Hz? Or perhaps it's something in how the keyboard decides to send the detected keys over the USB cable. Well, at least now I know: keyboards aren't created equal. This was a pretty cheap one, though it feels pretty solid.

Maybe I should try to learn a more disciplined, proper touch typing, instead of the 70% touch, 40% two-or-three-key harpeggios, and 10% backspace+retype I use now.

[1] http://en.wikipedia.org/wiki/Quantization_(music)

Upcoming?

This'll be a first status post, just so I post at least something today. I've written two or three drafts of a sort of status posts already, not quite right. The point would be to look ahead, and promise to post something. Or at least lay out some possible future titles, or give a direction. So that I may eventually come to do it.

The direction will be more of things not written much about so far: project & software architecture, philosophy (todo: find a better word), sociology, psychology, management (better word?).

I've made up some titles. They are in varying levels of drafting. Let's see...

Uncertainty Is Kryptonite
The Fog of Development
Recreational Downtime
When Is It Fun?
Estimagination
Findability in Project
No Code Is an Island (May become 'no language ...')
Remuneration Paradox
Is Happiness Necessary for Productivity?
Risk Sharing
Polyglot: Hot Or Not
Ideal Circumstances
Maximum Project Size
The Right Way

That's the new kind of posts. There'll probably also be some that are closer to the old kind...but better. Drafts:

List of Coding Habits I Have That Might Seem Odd
What I Worry About When Coding
How To Stop Worrying And Love Maven (Note: title is half lie, half wrong.)
The Eclipse Super-Flaw

Anyway, got to stop now. No promises. Later.

CSS Position Absolutely Got Better?!

Very odd, now "position: absolute" with both top and bottom (or right and left) works to affect layout! At least in Webkit browsers. Chrome and Safari.

Do you folks know what that means? Sane layout is now possible using CSS. And so, 2013 is looking to be a great new year already.

Unbelievable: Progress!

Chrome snapshot:

Imagine resizing and having the layout follow suit perfectly.

But not in Firefox. See how it falls -- wait for it -- flat:

Seems that the vertical layout just collapses.

Worth noting: I've never seen any documentation say that it should not work, i.e no wording like "but if you apply both left and right, or both top and bottom, results are undefined". But I haven't really read the spec.

So, when did this happen, recent Christmas?


------ update -----

Firefox 17: no change.
(Oh I wasn't quite up-to-date...)
Firefox 18: no change.