Updated 5-22-96
Updated 5-22-96
ISSN 1068-9109
Some of you may have been amused by the heading "The ABCs af OCR" in last month's article The Text Falls Mainly in the Twain. How did that "af" get in there? Easy -- the article was scanned then converted by OCR. Many of the errors were caught by the spell-checker. But the "af" was not because I had included the abbreviation "af" in the dictionary for use in an audio article, where "af" was used to denote "audio frequency." I've told you time and again to create specialty exception dictionaries for jargon and topic-specific words, so this won't happen to you. I did that but forgot to make sure the specialty exception dictionary was not used while checking the article. Why don't I listen to myself?
The other problem editors face is that the main articles usually have been proofread multiple times. But the last-minute stuff, like headlines, captions, tables of contents, and so on are lucky to get read even once. And it shows. Some of my favorites are (naturally) not my own creations: one was the headline that referred to Minier as our "Pubic Defender," and another stated "Three Killed, One Seriously."
These two errors show up different problems: the first is that many English words are still valid words when characters are dropped, added, or transposed. That's what limits spell-checkers to being only an aid for finding errors -- just because a spell-checker doesn't find any errors doesn't mean that there aren't any. Trail turns into trial or tail or train or frail, burn turns into bum (this one is especially difficult to deal with in OCR), and so on.
The second headline is in error only because it was too big for the allotted space: the word "Injured" didn't fit so was inadvertantly dropped. Only yesterday a client sent me a fax and asked how it looked. Previously we had had some character spacing problems so I carefully checked for that. I told her it looked fine, completely missing the fact that the whole top inch or so was missing, and the text just started in mid-sentence. Years ago I had spent considerable time dealing with various models of IBM electric typewriters. Somebody handed me a typewritten page and asked what I thought it was. I told them it looked like a Model C with a maladjusted typebar bucket. It turned out what they wanted me to comment on was the content of the document, not what produced it!
November 1995 Contents
Master Table of Contents
Leonora has a thing about the Emperor's new clothes: if you limit yourself to what the experts all tell you you will never make progress. In other words, be like the clear-eyed kid who could see for himself what was and wasn't so. She suggests that you try things and keep manipulating them until you like it pretty well, then show it around for constructive criticism. Don't show it too early, or the criticism will keep you from developing your own sense of what you want.
The real question when you show your work is "Did the piece do what you wanted it to?" Even if people say they don't like the design, if it accomplished its goal it is successful. And even if they did like the design, if it did not do what you wanted it to it is a failure.
November 1995 Contents
Master Table of Contents
Speaking of user interfaces, have you been bothered at all by the fact that within most applications you first select what you want to do it to, then select what you want to do (i.e., first select the text then select the font you want applied), but for dealing with the operating system you first select what you want to do then what you want to do it to (File, Open, then choose the file you want to open)? When I taught classes this dichotomy often confused computer neophytes: they could not predict when they should select the item first and when they should select the function first.
November 1995 Contents
Master Table of Contents
I took off the covers and didn't see much dust (not surprising because I had cleaned it only three months ago when it first became balky). So this time I took out a lot more screws and really got down into the guts of the thing. There was some dust on an item that I could not decide if it was optically important or not, so I cleaned that off (that particular spot had never been cleaned since I bought the drive, and was nearly sooty black). I peered, poked, and prodded but didn't find much else that looked promising, so I put the whole thing back together and tried it out. It worked. But that's no proof -- it often worked before, just not reliably. It's now 3 1/2 weeks later and it's still working, every day. Maybe it has nine lives like a cat. Even though it looks like I don't have to buy a new drive after all, I'm not sure if I'm happy or disappointed. But one thing is for sure -- the longer I put off the purchase the better off I'll be in tems of features per dollar.
November 1995 Contents
Master Table of Contents
This technique worked well for the various flavors of DOS (MS, DR, and IBM), networks, Windows (3.0, 3.1, 3.11, Windows for Workgroups), and applications (I have all versions of Corel so I can take a file created in CorelDraw 1.2 and upgrade it all the way to CD5, similar comments for WinWord, WordPerfect, and so on). However, OS/2 was not any real fun, although I eventually got it to work. But I didn't get to try out the HPFS (High Performance File System) which I would have liked to do. Win95 installed OK on the optical, but I needed to use Stacker in order to get enough space for Win95 and applications. At that time Stacker did not have a 32-bit version available, and the performance hit of running 16-bit Stacker under Win95 was just intolerable.
So what to do? I liked the concept, but the implementation I was using was running out of gas. I could use a Syquest or Bernoulli, but they are also about 200-270MB in size, not much different than the 230MB MO drive that I could buy. But I have had real bad crashes on Syquest drives and really didn't want to mess with them. The Bernoulli drives aren't too badly priced, but their cartridges cost so much more than optical cartridges that the MO drive becomes the economic choice after only a few cartridges. But 200-270MB just isn't enough for what I wanted to do, which is give Win95, Office95, Warp, and Netware 4.1 a good, thorough tryout. For this much more room was needed.
Noticing that 850MB EIDE drives were priced at $200 I decided that what I would do is reconfigure my computer from its current two IDE drives (a 114MB and 408MB) to a single drive, which I would put in a removable drawer. There are some side issues that I looked at long and hard, though.
One is that my computer is getting long in the tooth. Specifically, that means that it neither handles drives of over 540MB nor does it read the cylindars, heads, sectors info from the drive. That means that if I used EIDE drives I would have to use a disk drive BIOS extender to get access to anything over 540MB, and each time I exchanged the drives I would have to manually set the CMOS setup information for the drive (unless all drives were exactly the same -- but I knew they wouldn't be because some drives that I bought only a few months ago have been superseded by new models with more capacity, faster access, and lower prices). The alternative was to go for SCSI: the BIOS on the SCSI card would take care of all of this for me automatically. And I already have a SCSI card in the computer to run the CD-ROM, the Pioneer CD-ROM 6-disk changer, and the MO drive. So what's the problem? Just that the price for an 850MB SCSI drive was $85 more than the $200 for an EIDE drive of the same size. Add that up over the several drives I wanted (one for OS/2 Warp, one for Win95, and one for my normal production work) and the price difference was more than I wanted to pay. What's interesting is that if you drop down to 250MB drives, the price difference between SCSI and EIDE is insignificant, but at 800MB and above it is onerous.
So what I did was buy a $50 drawer and slide kit, and several more $20 drawers to with it, and two drives. The slide kit fits into a 5 1/4" drive bay, and the drawer holds a 3 1/2" drive. I put the current 408MB drive in one drawer (that's the current production system), a 850MB in another for Win95, and a 1.08GB in a third for OS/2 Warp and Netware 4.1. So far things are working well, but one thing I've noticed about these inexpensive drawers is they are all plastic, which means they can build up a destructive static charge. They also do not have any shock mounting, which means that we are depending on the shock mounts built into the drive. So I take great care to set the unused drawers on antistatic padding, and to ground myself before and while moving a drawer to or from the computer.
The comparison of costs is interesting: if I had bought a 230MO drive and three cartridges the cost would have been about $600. The Syquest and Bernoulli aren't much different. (Oops, that's wrong: Iomega was running a special on the Bernoulli with 5 cartridges for about $500. But it was still only 230MB.) But by going with the EIDE drives I needed to buy only two (because I could use the one I already had for the third) for $450 plus another $100 for the three drawers and one slide kit. For $550 I had more room, more speed, more reliability, and about the same cost as what one normally considers a "removable" drive. The only downside is that I have to waltz through the CMOS setup each time I change drives, but that's tolerable. Even though it takes me only a few minutes to change drives I typically do it only a few times a week. If the CMOS setup gets too tiresome I can buy a whole new motherboard and faster processor for less than the price penalty of three SCSI drives.
November 1995 Contents
Master Table of Contents
When I installed Office95 I chose the "Custom" install. It was a good thing, too, because some of the defaults were (in my opinion) peculiar. For instance, Excel95 has all the different spreadsheet converters built into it except one: Quattro Pro for Windows. To get that converter you have to go through the multi-step process of select the Microsoft Office for Windows95 Setup, Microsoft Excel, Spreadsheet Converters, and then either "Select All" or "Change Option" followed by select "Quattro Pro for Windows."
The Hyphenation and Thesaurus tools are selected by default, but Grammar isn't. Maybe Microsoft doesn't want to encourage its use but still get the advantage of being able to claim it as a feature in the inevitable product comparison articles.
A couple of other defaults I found interesting is that Graph is selected but the corresponding help file is not. And for graphics filters for file import PCX (PC Paintbrush), Corel Draw, and CompuServe GIF are all deselected. But each of these is very popular -- why aren't they selected by default? By the way, Microsoft claims the Corel Draw import filter works with CD3, CD4, and CD5. I haven't had the time to test this claim yet.
One of the things I really like is how Win95 treats the CD-ROM. When you put in a different disk Win95 looks at it and, if it is an installation disk (as in an upgrade of Win95, or the Office95 disk) it checks to see if you have already performed the upgrade or installation. If not, a dialog box appears telling you (in the case of an upgrade) that the programs on the disk are more current than what you have installed, and asks if you would you like to do the update now. This is very slick and makes the installation much easier than the time-honored File, Run, Setup (FRS) technique. Unfortunately it doesn't seem to be any better than FRS in avoiding breaking things that used to work just fine but don't after the upgrade or installation.
Snaggit 2.2 a review reprinted from the June 1995 issue of the Bytes of Las Vegas
PC911 a review reprinted from the Humbolt PCUG News Disk
First Aid for Windows a review reprinted from User Friendly, newsletter of the Los Angeles Computer Society
November 1995 Contents
Master Table of Contents