alice in intel-land

Somebody had to be the first one at work to get one of the new Intel-based iMacs, so I volunteered.

Thursday, February 23, 2006

Batting .500

As I've said before, there is certain software that people around here need to be able to run, preferably on their desktop computers. One expensive biggie is Matlab; people depend on it for data analysis. Not only can Matlab not be run on an Intel Mac, except in a no-graphics mode that really won't do; the Mathworks web site barely provides any official indication that they're planning a new release that will run on these computers (aside from a posting by one of the developers in a Matlab newsgroup).

Another widely-used program is Praat. It's freeware, and is used for acoustic analysis of speech (and so much more). It's truly amazing how much functionality is packed into a fairly small program. I'd downloaded the current version a week or two ago (then 4.4.06), and it worked wonderfully. From a user point of view, it's kind of irrelevant whether it's working in native Intel mode or in Rosetta (emulation) mode. The important thing is whether you can get your work done. And I was able to get some actual work done in Praat in this version. Today, I went back to the Praat web site (I know how often the developers update the program!) and, sure enough, they were up to version 4.4.11, which is the third version that's a Universal binary; that is, it'll run in native Intel mode. I'm going to have to get hold of some long sound files so I can really see how fast it is.

Meanwhile, I've been using Canvas to edit some bitmap images that need to be at a much higher resolution. It's a frustrating process, but the program itself is running just fine on Intel.

Friday, February 17, 2006

Moving to the dark side

In general, I disapprove greatly of system add-ons. On the one hand, they make it easier for a user to configure the computer the way he or she likes. On the other, however, they tend to break on system upgrades, sometimes in ways that are difficult, or at least confusing, to trouble-shoot.

But, the non-configurability of my Mighty Mouse was getting more and more annoying. I want my Back and Forward buttons for web browsing, dammit. And, as I was working on my map in Canvas, I was getting more and more frustrated by the seemingly random appearance of contextual menus, suggesting that my primary click was being interpreted as a right-click. I don't know whether this was a cause or an effect of the hand pain I was suffering from. But the lack of tactile feedback that would distinguish the two primary mouse buttons makes it an easy enough mistake to make. What's worse, one of the options on Canvas' contextual menu, at least in the mode I'm working in, was to reset the size of the area affected by the paint tools. So, instead of erasing one pixel at a time, as I intended, I would suddenly be erasing 49 pixels at a time.

So, I broke down and installed USB Overdrive, a neat little Preference Panel that provides greater control over mouse configuration. Even though there's a chance that it will break with OS upgrades (apparently, it broke with the upgrade from 10.4.4 to 10.4.5), the maintainer is quite responsive and the fixes are rapid. It's shareware/nagware; all features are available at all times, but if you pay the $20 fee you don't have to wait 10 seconds or so after you open it. This means that you can try it out before paying for it.

My primary goal was to get the Mighty Mouse side buttons to map to Back and Forward in my browsers. But I was delighted to find an additional option for application-specific button mappings. So, I disabled right-click specifically for Canvas. And I'm finding it a lot easier (though no less tedious) to edit my maps.

So now all I need to do is to pay for USB Overdrive.

Wednesday, February 15, 2006

Canvas vs Mighty Mouse

Today I installed the latest version of Canvas, version X. Canvas as an application has gone downhill since version 3, but people here use it, so we've got to check it out.

I've spent most of the afternoon working on some old maps that, coincidentally, I last worked on in Canvas 3 over 10 years ago. They're saved at 144dpi and the cross-hatching that codes crucial differences is too close together. Standards have changed, and, now that it looks like we might actually publish the paper that they're illustrating, I have to fix these things. Unfortunately, I can't find the original Canvas 3 files. What I have is the PICT files that were exported. This means that the only kind of editing I can do is bitmap editing. That is, I have to manually erase half the lines in each cross-hatched area. And, yes, that is exactly as tedious as it sounds! Once I'm done with that, I'll have to change the dpi setting up to, probably, 600 and smooth all the curved lines. This will be at least as tedious as fixing the cross-hatching. I'd hate to have to do this as part of ordinary work-flow, but, as it's a one-off (well, there are three maps), that's fine.

I had three problems, though. As with the other problems I've had with the new iMac, it's not clear to me what part of the technology is causing the problems. That is, I've encountered them working with an Intel iMac, but it's entirely possible that I'm just trying things that I haven't previously tried with G3- or G4-based machines, and that, if I had, I would have had the same problems.

The first problem was opening my old maps. The old creator code is recognized, but opening them wasn't so easy. I have the hard drive on my older iMac mounted as a volume on the new iMac, and it shows up in the Finder and any mini Finders windows. Yesterday, I had no problem opening files on it from Praat on the new iMac. But when I tried to open the Canvas files that way, I ended up with the spinning beachball and had to force-quit the program. Twice. Then, I dragged copies of the maps onto the desktop of the new iMac. They show up with CV3 (Canvas 3, I assume) icons, and open fine if I double click.

The second problem has to do with the Mighty Mouse. When I first started exploring what the various buttons do, I noticed on the Keyboard and Mouse Preference Panel a checkbox to enable horizontal scrolling. I couldn't imagine why anyone would want to do that, but I checked it anyway. Well, editing these maps, it became perfectly clear to me why you'd want horizontal scrolling. In order to easily perform the digital manipulations I wanted to perform, I was working at 1600% magnification. So, of course I want to scroll horizontally. There's only one problem: it doesn't work. Sure, I can move the scroll button left-to-right or right-to-left. But the cursor won't move horizontally. What's worse, attempting to scroll horizontally produced unexpected vertical scrolling. I have no idea whether this doesn't work because of OS 10.4.4, the Intel architecture, or the way that Canvas is programmed; it doesn't work the way it should. It doesn't make Canvas not work; but it makes my mousing hand hurt when I'm working in Canvas. (It does work they way it should in this composition window, not that there's a whole lot of horizontal scrolling necessary here!)

The third problem is printing. The first time I printed my map, two edges were cut off. The right edge is cut off c. 3/4" from the right edge of the page, but there's a good 2 inch margin at the bottom and the map is still cut off. This is a printing problem only; the maps look fine on the screen. The first time this happened, I went in to Page Setup and found the option to scale the picture to fit on a single page. This setting made no difference at all; the two copies of the map that I printed are the same size and are truncated in the same way. Looking at it again, though, it appears that Canvas somehow decided that I had a custom paper size of 7.5 x 10.5 (instead of 8.5 x 11). I have no idea how that happened!

Tuesday, February 14, 2006

Flashiness and all that Java

One of the things I've read about the new Macs is that there may be problems with Java, Javascript, and Flash plugins. This is supposed to be the reason that Matlab graphics won't work at all. The best tests of the new system that I know of therefore involve games and add-ins that make use of these technologies. So, I've taken this as an excuse to play Sudoku. You have to have Javascript enabled for pencil marks (the ability to indicate tentative choices) to work properly. On both my iMac G4 at the lab and my iMac G3 at home, the option works, but the implementation is a bit choppy; numbers don't appear exactly where in the square you expect them to, and, if you have more than one number in a square, there's a funky little extra box around it. In contrast, on the Intel iMac, the implementation is exactly what you'd expect. Numbers are the size you expect, and there's no funky little box. So, there's a plus.

In addition, the Flash-based game Net works as expected in Safari and FireFox. I really can't justify any more game-playing at work, so I'll have to look out for other opportunities to test out Flash, Java, and JavaScript. I suspect that a lot of stuff that has been reported to break has been pushing the technology in unexpected ways. But, we'll see.

Wednesday, February 08, 2006

Reset

I tried to right-click for a contextual menu this morning, and it didn't work. When I went in to System Preferences, it turned out that my Mighty Mouse had somehow been reset so that right-click and left-click did the same thing. I've fixed it, but it's something to look out for, lest it happen again.

I just noticed that the little wheel scrolls horizontally as well as vertically. That's pretty cool.

Finding out for myself

As part of learning about this new-style iMac, I've been following along with other people's experiences. But what I'm learning is just how large the grains of salt are that I need in interpreting that experience.

Two cases in point:

1) Something I read yesterday (I forget whether it was on one of the comp.sys.mac.* newsgroups or Macfixit/Macintouch) suggested that an Intel iMac on a mixed network might have trouble finding and connecting with other machines on the network. So, I immediately hit the Connect to Server command on the Finder's Go menu, and browsed the network. The display was a little different from what I'm used to. (Of course, checking the old iMac reveals that the differences are 10.4, not Intel-based.) I get a listing of all of the Windows workgroups on our network. Also on the list is an item "My Network", which contains all of the Macs on the network. The various lists are populated much more quickly than on my old iMac, and the login connection to the old iMac is nearly instantaneous; and I had no trouble copying some installer files I needed.

I have no doubt that the original poster was having difficulties connecting to other machines on his or her LAN. But, clearly, those difficulties have to do with local conditions (the configuration of either the computer or of the LAN). So far, I've had no difficulty with the network.

2) A friend on a mailing list who also has a new iMac reported difficulties with Office.X, even though it's widely reported that the various Office apps run just fine under Rosetta. Back and forth conversation on the mailing list elicited the information that he hadn't applied the various update patches, in particular the 10.1.2 Updater. Once he did, Office.X applications just worked. So, I pulled out an Office.X CD and installed it, together with the three Updaters that I'd saved on my old iMac (see point 1 above). As I'd expected, it works just fine, but I still need to pull in a long, complicated document to see how quickly I can scroll through it. Also in the course of our discussion, one of the participants observed that the 10.1.2 Updater has been pulled from the Microsoft web site. Well, yeah, it has, but that's because it, along with the 10.1.4 and 10.1.5 Updaters, both of which I'd saved, has been rolled into, ta-da, the 10.1.6 Updater!

Friday, February 03, 2006

Unprintable no more

It looks like my problems printing to the multifunction printer/coper/fax are OS rather than hardware related. While obsessing over yesterday's problems over dinner last night, it occurred to me that the computers on which printing worked with the HP gimp PPD were computers that had been upgraded from Panther to Tiger. Since we don't generally do "clean" installs on upgrades, it's possible that some older drivers were lying around that wouldn't be available on machines that came with 10.4 on them. Thus, the iBook. It was purchased last summer, and is currently running 10.4.3 (only because I haven't gotten around to upgrading to 10.4.4).

And, lo and behold, it was giving me the same problems. I could print just fine from my old iMac, but print jobs stalled from the iMac, when the printer was added in exactly the same way.

And then I remembered the CUPS link! When I looked at the listings for the printer on the old iMac and on the iBook, they had different Device URIs. I manually changed the device URI on the iBook to "socket://xx.xx.xx.xx/?bidi" just as it appeared on the older iMac, and clicked on test page. The test page printed just fine.

So, the next thing to do was go back to the Intel iMac. I modified the printer description that I'd defined using the Print Center, printed a test page, and Bob's Your Uncle!

I can print, and I can even print double-sided. Life is good.

Thursday, February 02, 2006

Unprintable

The laboratory has two types of networked printers. We have a variety of HP Laser Jets, some publicly available and some not. The older ones are reachable only by AppleTalk, but the newer ones have IP numbers and/or announce themselves by Bonjour/Rendezvous. We also have three multifunction machines. Two are photocopier/printers, and the third functions as a fax also. The company that provides them rebadges them; two are made by Toshiba and one by Canon. Printing to the Toshibas, including the one closest to my desk, has always been an iffy proposition, especially from a Macintosh. But, generally, it's possible to print, using the Generic Postscript driver. Before I upgraded to Tiger, I was able to define a printer queue based on its IP number, and use a PPD file that allowed me to do things like duplex. But since, the upgrade, that hasn't worked. A colleague, however, has managed to print to it and access these features using an HP driver, but I haven't been able to make that work.

For the past hour, I've been trying all sorts of permutations (a lot of HP LaserJet 5 PPDs are available), and have been unable to print to the copier/printer. At best, I've been getting semi-informative error messages like "unable to get printer status (server-error-operation-not-supported)." Using the CUPS HPLaserJet 4Si Gimp PPD that my colleague has no problem with gives me "Printer busy; will retry in 10 seconds..."... Hmmm...wonder if it's a firewall problem. I'll have to wrestle with that tomorrow. Fortunately, I had no difficulty printing to the closest true HP printer.

UPDATE: Defining the printer via AppleTalk rather than IP, but using the HP PPD gives me still a different set of errors. Essentially, printing stalls at 24% of the first page. In contrast, the same printer setup from my old iMac works fine. Something's different here.

SPSS and Mighty Mouse

One of the web sites I browsed yesterday expressed surprise that SPSS 11.04 worked on an Intel-based iMac, given how late SPSS usually is to the dance; many of our users had to delay upgrading to Tiger, or even purchasing an upgraded computer, because of how long it took SPSS to issue a patch for Tiger.

So, this morning, I pulled out my sheet of license codes and my 11.04 CD, and installed SPSS. The installation went smoothly, well as smoothly as on older Macs; the license codes we were issued by Yale (and that my Yale contact insists are correct) indicate that our license expires on 4/1/06 rather than 12/1/06, as they should; but I'll cross that bridge when I come to it!).

I was able to run various analyses in the Tutorial, but will need to borrow an actual, large data set from a colleague to see how fast it is.

One thing I noted is that the scroll-wheel on the mouse doesn't work within SPSS. I was forced to click in the scroll-bar. I don't know if this can be fixed with something like USBOverDrive, or whether it has to do with the whole Rosetta vs. native thing. In any case, I haven't noticed it before.

But that brings me to the Mighty Mouse. I really dislike it. I've been using Logitech multi-button mice for a number of years, so I'm used to a decent-sized scroll-wheel. The little dimple on the Mighty Mouse is just too small, and is reminiscent of the obscene dimple used for "mousing" on some Windows laptops. I would have expected, at the very least, that right-clicking would have been enabled by default. It wasn't. I had to get into the Keyboard and Mouse Preference Panel to enable it. No biggie. What's worse is that the little buttons on the side, which it's almost impossible to avoid clicking when you move the mouse, are pre-configured to invoke Expose. So at the least movement of the mouse, the window you're attempting to navigate in escapes from you, and you have to chase it down. You can turn off this behavior using the Preference Panel, but you can't program the buttons to do something more useful. In contrast, the homologous buttons on one of my Logitech mice map the the forward and back buttons on Safari (and, I think, Firefox). Likewise, the scroll-wheel, when pressed rather than scrolled, invokes Dashboard. This isn't as big a nuisance, as I haven't pressed it accidentally, yet.

To be fair, some of this behavior may come from Tiger per se, rather than the Intel architecture. I've noticed some similar problems with the wireless mouse I use with my iBook running Tiger, especially random and inconvenient invocation of Expose, but I haven't tried to reconfigure the buttons. Yet.

Wednesday, February 01, 2006

I code, You code, Unicode

In order to set the iMac up for regular use, I had to start downloading.

There are several phonetics fonts that are more-or-less standard in the field, so I had to download them from the SIL web site. The download was excruciatingly slow, causing me to wonder about the vaunted speed of the new computer. However, other downloads (Thunderbird, MTNewsWatcher, and StuffitExpander) were appropriately speedy. In addition to the fonts, I downloaded the SIL IPA keyboard layout, as using the Character Palette for entry of Unicode characters just doesn't work when you're using a lot of them. Long-time users of SIL's pre-Unicode fonts tended to know where various characters, such as ə live (in this case, it's shift-option-e), so having a keyboard layout that honors that arrangement is quite handy. After putting the fonts and the keyboard layout where they belong (/Library/Fonts and /Library/Keyboard Layouts, respectively), I restarted the computer (probably unnecessarily) and fired up Pages, or should I say, Pages 2.0? In any case, there are a lot more built-in templates than in the original version, including envelopes. But I used the boring blank template (I'm not sure that I would even call it a template, but that's just me). All the keyboard switching and font switching, unsurprisingly, works just fine.)

(In the process of preparing this entry, I discovered that the current version of Safari (2.0.3) doesn't support the various tools Blogger has available for easy composition of HTML-formatted entries. Oh well.)