alexpgp: (Computing)
I've looked around on the Web to see if there's any good info about a problem I've been having, in which double-clicked applications get to a point where some number of kilobytes—typically 104 KB, but sometimes 106 KB or 108 KB—of memory have been allocated, and then... nothing. The amount of memory used by the application does not increase (so naturally, it doesn't appear on the screen), and double-clicking the application again will start another copy of the application, which gets to the same point and stalls.

Most of the chatter I've found is associated with people who run into this problem when attempting to run Firefox, but I don't run Firefox. Despite that, over the past couple of days, I've run into the problem soon after my machine has finished its startup routine, which is not terribly encouraging if you consider that the only "cure" I've found to this problem is to reboot the machine. I do have better things to do.

Then something I caught "out of the corner of my screen" while browsing potential solutions turned out to be a huge help and points directly at Windows 7 as the culprit (as far as I can tell).

Programs can successfully be invoked from a command prompt. I therefore conclude that there is a serious problem with the mechanism by which the OS turns a double-click into a running application. (FWIW, hitting return after highlighting a file in a directory window doesn't work, either).

More research is required, but at least I have a workaround that won't force me to reboot if this phenomenon raises its ugly head at a particularly inconvenient time!

UPDATE (and data point): Thunderbird allows me to open email attachments by right-clicking them, and doing so on a Word file has just successfully opened an attached file in Word. Therefore, an application can open another application, too. Just can't double-click on it, is all...
alexpgp: (Computing)
...wouldn't bother me at all, if I had flowers on the wall.

But various Windows problems that keep recurring with no rhyme or reason (and for which no good information appears to be available on the Web) are really starting to disturb my equilibrium.

The most common problem has been my system going into an infinite loop waiting for something to happen with respect to the System Event Notification Service. And although having to reboot through an offer to do so in Safe Mode is not a big deal, it's an annoyance, plus there may be things that are not getting done—things that would occur after the SENS was taken care of—that might be important.

Most recently—about a half hour ago—another problem cropped up which has occurred only a couple of times but is most annoying: Windows 7 suddenly refuses to load more than 104 KB or 108 KB of an application before Going On To Other Things™, Typically, what will happen in such circumstances is that I double-click one or more additional times on an application's icon (thinking perhaps I had failed to do so properly the previous times), and only after a while does it occur to me to check out the Task Manager, where I will see however many instances of the app in the process list, each staking out its own 100 KB or so of memory.

So, seeing as it's Sunday (and that we've just gone over to Daylight Savings), I decided to see what wonders might be lurking in Windows Update.

Who knows? It might help, and stranger things have happened.

alexpgp: (Computing)
My old Hawking HWPS1UG wireless print server was a pretty useful piece of hardware once it was properly configured, and there lay the rub. I recall working through the various steps in the configuration process a dozen times or more before somehow, something "clicked" and the unit started working.

For those unfamiliar with what a wireless print server does, it's basically a box that links wirelessly to your home network, from one "end," and to a printer, from the other "end," and thus allows you to print stuff without there being an actual wire connection between your computer and the printer. Even better, multiple computers on the same network can use the same server, allowing print "jobs" to originate from more than just one computer.

The problem with the Hawking, besides it finicky nature, is that its idea of network security extends to WEP, which is no longer a secure way of running a home network. So I went out and got a Trendnet server and had a go at installing it today. I learned a lot, too.

For example, if one's router supports something called WPS, then you don't have to worry about how a new WPS-compliant device (like the Trendnet) gets connected to the network (which normally requires a password for a successful connection).

But you do have to have WPS enabled on the router, which I didn't, and I needed the router's password to enable WPS. As it turns out—in a sort of good news, bad news sort of way—I actually had the password close at hand, but it was actually accessible to anyone who could gain physical access to my house. (By default, but not any longer. :^)

I was not terribly thrilled when the Trendnet software was reported as wanting to run every time my system booted. This had not been the case with the Hawking (when I used the Hawking server, I was directing print jobs to a special version of my printer's driver). With the Trendnet, when the software runs, it automatically "connects" to my printer in such a way that when a print job is sent to my printer, the output from the standard printer driver is apparently intercepted by the Trendnet software behind the scenes and quietly relayed to the printer.

In any event, the thing works, so as long as the additional process doesn't eat up too many cycles, I guess I can live with it.

alexpgp: (Computing)
At the excellent suggestion of LJ friend [ profile] eastexpert, I downloaded and installed something called "Core Temp," which is a small application that tracks CPU core temperatures.

Back in 2001, my work laptop's fan stopped working, which caused the CPU to overheat to the point where it suffered irreparable damage, and although replacement of the malfunctioning fan and the damaged CPU was covered by the manufacturer's warranty, the episode—which involved sending the computer away for several days—did put a dent in my productivity.

When I fired up the "Core Temp" application last night, the temperatures of the CPU cores stood at about 55°C, well below the theoretical maximum of 105°C. Since then, I've read posts on various forums that suggest a temperature in the 50s—while not terribly high—is perhaps a bit higher than it should be. Indeed, since booting my computer this morning, the core temperatures have not budged from 25°C (although I have done nothing to change the computer's physical configuration since yesterday).

One hypothesis I am entertaining about the cause of the elevated temperature involves a process that gets started each time I boot my machine and which I have noted at the top of my process list when said list is arranged to show the most CPU-greedy processes. It's too soon to tell if there is any cause-and-effect, but I have experienced no freeze-ups since killing said process. (Deleting/uninstalling the process from my machine does not seem feasible, at present, as I actually do need to run the program from time to time. I may re-evaluate just how badly I need to run the program if I experience no freeze-ups over a significant time period.)

And so, for the foreseeable future, vigilance will be the watchword as I seek to keep my machine up and running.

alexpgp: (Barcode)
Some time ago, I had noted that my LJ layout (Flexible Squares) had suddenly begun to display text posts in such a way as not to "wrap" around my userpic.

Then I noticed that, despite this behavior, the text of comments did wrap, and that got me to thinking—go figure!—about another change that took place at the time that text stopped wrapping, and that had to do with photographs being "cut" to column width if they were wider than what the style allows for "normal" column width (not optimum behavior, but it is an annoyance I can live with).

Nothing I did with the CSS code for ".entry" had much of an effect, and then I noticed the code for ".entry_text" and ".entry_text .ljtags":
.entry_text { overflow: hidden; }
.entry_text .ljtags { clear: both; }
The first, I reasoned, is responsible for cutting photos off at the right-hand margin, and according to a Google search, the value of "both" for the "clear" property basically keeps floating elements from appearing on the left and right sides of an element (e.g., the content of an entry).

So, clicking on "Customize" and then "Customize your theme" under where my current theme was shown, and then selecting "Custom CSS", I scrolled down the page to the "Custom stylesheet" text box, entered:
.entry_text { overflow: visible; }
.entry_text .ljtags { clear: none; }
clicked on "Save Changes", reloaded my journal, and BAM!, text was shown wrapped the way it used to!

Between this, coming back up to speed in GNU emacs, and learning about org mode and BBDB, I'm feeling computationally accomplished!

In other news, invoices are caught up to date. As far as the rest of the year is concerned, it's difficult to predict what the next couple of weeks will bring. On the one hand, it's the holiday season, which might lead one to think work will be slim, but historically, that's not the case (at least not for me).

Looking at the overall numbers, I completed 6 assignments in the last 15 days of 2009, and 13 in the last 15 days of 2010. Part of the explanation for all the work lies with how people about to take time off want documents to be translated in time for when they get back. (That explains why I get so much weekend work. It's an occupational hazard for translators, but the money earned pays the bills just as well.)

alexpgp: (Computing)
I called Acer yesterday to get some clarity on the failed hard drive, and was stymied from the very beginning. Their automated system requires you to "say or enter" one of two numbers, the first of which I could not find, and the second of which consists of 20+ numbers and letters.

After a few tries, the system finally said it had understood what I said, and then proceeded to read back a number where "B" was replaced with some other letter, and then again with some other letter.

At this point, I began to repeat "representative!" rather loudly, whereupon the system basically told me that I would have to go online to get support, and then hung up on me. I tried calling again, but apparently the system uses Caller ID to cut the transaction short, and I was again told to go online.

In the end, I was able to engage in a chat session with someone from Acer, whose position basically was that a hard drive could not fail the way mine had (loss of partition table upon power-down) unless the system had been improperly shut down. And while I admit to having pressed the on/off button for the requisite four seconds to shut the system off instead of doing it through the network, it occurs to me that four seconds is plenty of time for the hardware to move the hard drive heads out of harm's way before power is lost.

Anyway, based on my experience—numerous computers over the years that have lost power under more egregious conditions resulting in data loss, but never in a trashed disk—I now look at the easyStore hardware and wonder whether it's worth trying to recover the system at all.

alexpgp: (Z-z-z-z-z-z-z-z-z)
I took the boot drive out of my home server and hooked it up to one of those universal doohickeys that provide data and power connections to a USB interface, effectively turning such a disk into an external USB drive.

After going hmmmmmm—BURP! about a dozen times, Windows quietly gave up trying to do anything further with the drive.

I then plugged the getup into a Linux box, with much the same results, except that in Linux, one can type
fdisk -l
at the command line and get some insight into the problem, which in my case was summarized by the line
Disk /dev/sdb doesn't have a valid partition table
This sort of made sense, considering the cavalier manner in which Windows had dealt with the drive. That said, the news did not make me a happy computer camper.

Searching for a solution to my problem on the Internet turned up quite a number of cases where 1-TB drives from Western Digital go belly up like this (or perhaps that's just my perception).

In any event, I plugged the drive back into the Windows box to confirm it doesn't show up in Disk Management (a couple of sites suggested solutions if the drive did show up as present but unformatted), so now I'm 980,000 sectors into a 4,294,967,294-sector scan of the drive using a tool called GetDataBack from Runtime Software.

Just how long this is going to take is not clear. When the scan started, the time was around 21 hours. Right now, it's about 18 hours. As I understand it, the boot drive on a WHS system has two partitions by default: one for the system and the other for data. I am interested in the latter, and as there is nothing else to do until the software has completed its analysis, I shall remain optimistic.

Only 99.9967% of the disk left to scan!


UPDATE: Getting at the data on the drive is not a critical issue. Upgrading my netbook's hard drive this weekend is. Throw in the fact that I started the scan with everything plugged into an unprotected power strip (i.e., one not plugged into a UPS) and that we've had momentary blips in power due to recent thunderstorms, and it's clear I should do the scan later.


alexpgp: (Default)

September 2017

      1 2
34 5 6 7 8 9
1011 12 1314 15 16
17 18 19 20 2122 23


RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 24th, 2017 12:16 pm
Powered by Dreamwidth Studios