The comments problem I'd been having? The one that triggered a semi-painful blog rebuild? Turns out it was all due to a <MTCloseComments> lurking in the Category archive template.

I figure in order to keep the whole mess worthwhile, I'll just leave that in the old blog as a nice clean demarkation between where comments Go and where they Don't.

I'm so fucking awesome it hurts.

Having skimmed the plugin documentation a bit more, it looks like it keys off of the number of days displayed in the default index. The thing is, this variable is set in the main blog config, and can be freely ignored by template implementation.

Oops.

This data structure gets the same listing results under 10.4.11 and 10.5.2, from different Macs and different user accounts with different degrees of fresh/updated OS installs. The directories list in the proper order in the command line and on Windows.

 

What. The. FUCK?!, I ask you. This is almost as frustrating as Safari 3 dropping the sidebar from the file dialogue (on 10.4, which I'm still using as my primary workstation since the install of 10.5 onto the video box completely destroyed elements of my Final Cut Studio install and I'd rather stick with something that works than spend the time to figure out why the new thing doesn't), only with the added advantage of having persisted across multiple revisions of the OS.

I'd have complained sooner, but had I done so my grumping would have encompassed only 10.4.X. And it's not like I have a day-to-day need for the contents of the 2005 directory.

This doesn't happen with my ATC pages, fortunately - they're all named p_$pagenumber for The Dualist, v_$pagenumber for Transitional Voices, and d_$pagenumber for DCR. They all list just fine. The image directories for 2006 list just fine, as do the directories for 2007 and 2008. If I change the 2005 directories to something like 01_200501 02_200502 etc, then they list correctly. It's only 2005 that lists wonky, and it's not just the user-generated directories - it also lists incorrectly in the /Users/$user/Pictures/iPhoto Library/Data/2005 directory, which is generated by iPhoto (an application that phenomenally fails to meet my image archiving needs, but that's another story).

What could 2005 ever possibly have done to piss you off, Apple?

Seriously. Inquiring minds want to know.

Update : Turns out it's not a bug, it's a FEEJUR.

<@ejp> what's the issue with that order?
< solios> it's WRONG is what
<@ejp> how?
< solios> 200504 does NOT come after 20080-anything.
<@ejp> 20050427 does. it's a bigger number.
< solios> likewise, 200510 should come AFTER 200509, not before 200504.
<@ejp> it's alphnumeric sorting, not years
< solios> ...
< solios> if it is it's only for 2005. :P
<@ejp> no, your 2005 directories have more digits.
< solios> I changed 20050427-0429 to 200504 and it listed correctly but the fact is it still ought to be first regardless.
<@ejp> no
< solios> no?
< solios> well why not. :P
<@ejp> 2005010101 is a bigger number than 20089, mathmatically speaking.
<@ejp> finder has no way of knowing it's a year.
<@rjbs> maybe you could use a resource fork
< solios> ejp: neither does the command line or winscp but they still get it right.
<@ejp> yeah, some things use more complex sorts.
<@ejp> *shrug*
< solios> so you're saying finder's giving me bullshit results because it's "smarter" ?
< solios> it makes my brain itch! @_@
<@ejp> na, it's slightly less smart.
<@ejp> or at least does less stuff
< mdxi> that is a weird sorting order, but i see its own, retarded, internal logic
< mdxi> it's not asciibetical because _ is after all numerals
< mdxi> but sorts ahead of "nothing" here
< mdxi> oh, as does hyphen
< mdxi> lesson is be consistent, i reckon
< mdxi> because apple hates you
< solios> MADNESS!
<@rjbs> you'd be better off with yyyy/mm/topic
< solios> 2005 is subsorted like that because it's fully half of my camera images. I suppose I can clean things up so it'll list correctly, but it doesn't change the problem. It just avoids it.
<@rjbs> because everything will sor tthat the same
<@rjbs> use File::LinkTree::Builder :)
< solios> hm.
* solios files this under 'work he shouldn't have to do' and goes back to sorting the Navy/turnpike pics for posting.

Way to go, Apple.

Update : "Fixed" this by altering my naming convention and giving 2005 a thorough cleaning. This doesn't solve what I still see as a bug, it just ignores the problem.

For future use:

10:41 <@ejp> 'cat /proc/net/ip_conntrack | wc -l' will tell you how many connections are active

Disclaimer : I'm a rank amateur when it comes to network mechanics, both in theory and practice. The fact that the problem at hand responds to shotgun debugging doesn't mean I'm necessarily doing anything right, or that I even fully comprehend the options. Frankly, my brain has other problems to work on, and I'm willing to settle for being the only artist I know who can follow a conversation about linux.

====

One of the nice things about being able to ssh to gridlock from work is I can do a little thing that would be a real pain in the ass to do at home - I can drop the interface between gridlock and the home network and See What Happens.

Lately, this is what happens:

A thing I miss about Jen - once I tweaked her limewire preferences, my pipe got useable and stayed that way. Randy, conversely, has been sucking on the thing so damned hard it's a wonder that gridlock hasn't exploded from the strain*. We're talking "I have to physically unplug his ethernet cable to get any degree of downstream" kinds of sucking.

If only the Linksys WAP I've stuck between him and the network could limit downstream as well as upstream :P. I'll experiment with capping flow control when I get home today - hopefully that'll solve the problem**.

Seems like I may be in for a modern version of the "war of the modems" I had with Brooke (my first roommate in the current house), with the difference being it's my money and hardware, and I have more control over the situation now than I ever did then. I really hate the idea of cutting off or throttling internet use. It makes me feel like a complete asshole. But in my mind, there's a difference between use of a free service and brutal assrape of a free service - the former isn't a problem. The latter obviously is, or I wouldn't be ruminating about it on teh intarwebs. I doubly wouldn't be ruminating if my roommate were easier to get ahold of.

13 hours later - hey, looky! : It was the router all along. A hard reset fixed the pipe completely. That's how it always seems to turn out. Kudos to ejp for the troubleshoot (and for adding bwm). All of my above gripes re : the roommate's bandwidth usage still holds, as a high volume of connectivity always seems to make the thing crap out.

20080411 : After adjusting his client and exchanging words (through paper), the problem has yet to resurface. Thank gid.

* It has, actually. And so has the DSL router. An enormous volume of connections has a tendency to make both of them sick. Sick to the point that my connection actually dies. Dies as in "no noise on the phone line for a few minutes." Dies as in "too many connections will cause the bandwidth to collapse from 150 KB/s to 40 or 20 KB/s and stay there until the router is manually reset." Or worse, just trail off to b/s and then nothing. Initially, I thought this was a hardware problem, but after extensive amateur debugging, symptoms and circumstances and evidence all point to user abuse.

** The problem is that we're on diametrically opposed schedules, so I can't exactly knock on his door and yell STOP RAPING TEH INTARWEBS PLZ. I've laid eyes on him twice since Jen left for RTC, which means indirect methods are going to be necessary as a stopgap until I can actually talk to him.

So. Somewhere between the last time I needed to do subtitling at work and today - when I'm doing it again - there was a Quicktime update. An update that induced some pretty serious gayness, as I've illustrated above. The player used to display the FPS of a file using the timecode of the file itself - 29.97 in this case - and now it displays it in base 99. For some fucking reason. 00:53.85 in the player is actually 00:53.28 (or thereabouts) in the editing and compositing software.

I don't know what manner of asswhole thought this was a good idea, but they deserve to be punched for making my job a hell of a lot more tedious.

My options are to keep Calculator.app out for conversions, or to switch to using Final Cut Pro in place of the Quicktime Player, since FCP actually displays the frigging FPS properly.

Update : Turns out FCP works exceedingly well in this role, better than the Quicktime Player ever did. However, Teh Ghey has spread to After Effects, which repeatedly runs out of memory on the G5 (which has 2048), while functioning Just Fine (and, in fact, a hell of a lot faster) on the G4/733 (which has 1024). Go figure.

I've been having some Serious Problems with Photoshop on my G4 Digital Audio. The major issue is that the machine's a complete slug running 5.5 in Classic on OS X, though this week that age-old gripe has taken a backseat to the fact that Photoshop CS is crashing like a crackwhore trying to deliver a blowjob in the midst of withdrawal. This started last night as a repeated load-then-crash on a document I was working on, which I worked around by loading the app first, then loading the file. I was able to work on it easily for some time, until it came time to convert the document from Greyscale to RGB color. CS crashed. Then it crashed again. Then it locked the machine. Reboot, retry, lock. Reboot. Do the conversion and flat colors in Photoshop 5.5 running in Classic - the RGB convert occured instantly - and save off. No problems.

Try the file in CS again this morning and the machine hard-locks again. Good thing the G4DA has a really convenient reset button - none of this "hold down power until the machine feels like Stopping" crap. This is exceedingly Odd behaviour for CS, which has functioned fairly well whenever I've needed it, which isn't often. I'm assuming this has something to do with the processor upgrade I administered in January, but I can't imagine what, as everything else is fine. The last time everything went crash-happy (halo, wolfenstein, photoshop, Classic, everything-but-the-OS), I straightened things out by flipping the dip switches down to 1.33 ghz from the Factory Default of 1.4. Still better than 733, especially for Photoshop.... though I had to remove the Altivec Core for Photoshop 5.5 to prevent it from Disappearing from memory whenever I ran anything that had been G4-accellerated. Gayness.

This recent crash-happiness has me thinking about doing a full ground-up reinstall and rebuild of Minerva. It also had me thinking about buying a Mac Mini, despite the fact it's a single-head machine and the same speed as Minerva, as I'd rather have an equivalent box that fucking works than a beefed out piece of kit that doesn't, and the Mini is the only thing in my price range. I don't have - and couldn't possibly aquire - the two grand I'd need for a G5 (more like six grand if I'm going to do it right, which would be a g5, at least 2g of ram, scanner, monitor, tablet, and a PC to plug into the other KVM slot for 3d purposes).

Right on cue, Apple Intels the Mini, which drops it right out of the running as an option. I need Classic. As annoying as this is, certain quirks in Photoshop CS handling - currently the fact that it kicks Minerva in the balls every time I try to do something with it - have me pretty irked about the future direction of the Mac platform.