Archive for the 'Linux' tag

I never minded stuff like this when the ebooks I was reading were being produced by volunteer labor, but now that I'm paying real money for them, it really irritates me.

I'm reading The Confusion, volume 2 of The Baroque Cycle by Neal Stephenson, which I purchased from fictionwise.com.

The chapter I'm reading takes place at a castle in Germany, and I believe Stephenson refers to it by the german word Schloß. However, the producers of the ebook got the code for the German double s wrong, and so instead of a Schloß, the ebook keeps talking about a Schloé. (Html entity 233 instead of 223.)

This book is published by "William Morrow, An Imprint of Harper Collins Publishers" and I'm sure there are lots of people who work for that organization who could spot a typo that bad and that consistent, so I can only conclude that none of these proofreaders was asked to look at the book after the people who converted the text to the epub format were through. And that the people who did the conversion aren't good proofreaders.

Of course, this would be even more irritating if I weren't running the illegal script that turns the ereader format back into html, which I can edit with emacs.

I wrote previously about my efforts to find a browser to replace Firefox 3.0, which has major memory leaks and takes over the sound system.

I seem to have settled on Firefox 3.5 for the moment.

I still like the interface on chromium-browser from google, but the linux version was too incomplete, and so I had to keep a firefox browser going in addition. In addition, because it was undergoing such rapid development, I was having to restart it every day, which is a nuisance.

They have actually gotten flash working, so you could use it to watch youtube videos, but I'm still not able to publish my daily blog entry. And they frequently have problems with the interface with X windows, so that you can't move or view the window the way you expect.

Firefox 3.5 still has some memory leakage, and of course when it's using 13% of my current 8G, that would have been more than 100% of the old 1G system. But it takes at least a week to get to be a nuisance on the current system, and by then I've usually had to restart it so that I can listen to the MIDI files when I'm transcribing music. (This is a point for Chromium; I can play midi files from the command line even if I've just listened to music on Chromium.)

Another major advantage of Firefox over chromium is that there are all those plugins, including one that lets you use emacs to edit text fields, and the one that lets you share your bookmarks between all your computers.

I've heard people complain about problems with Firefox 3.5, but the only one I've hit is that my bank site complains that I'm using an untested browser, but then it lets me do my banking anyway.

So for now, I'm putting up with Firefox 3.5, but I'll let you know if chromium grows up enough to be worth another shot.

I went to a Boston Linux and Unix Users Group (BLU) meeting last night to hear a talk on MythTV by one of its developers (Jarod Wilson who works for Redhat).

Some points of interest about real world MythTV use:

• Recording from a cable box is more haphazard than you would wish -- the most reliable way to record anything you've paid to be able to watch is to get the Hauppage HD-PVR which lets you plug in the composite video cables from the cable box and use an IR blaster to change the channels. Modern cable boxes have a firewire output, which should let you both record digitally and change the channels, but it's fairly haphazard what channels your cable company will let you see unencrypted on the firewire output. Also, the HD-PVR will allegedly record in either 720p or 1080i, but there are some issues with the linux drivers for interlaced video, so you're currently safer sticking to 720p.
• Most of the USB remote control boxes on the market, including the HP branded one that I inherited from Bonnie, are essentially the same as the Windows Media Center one.
• If you're setting up a filesystem partition for mythtv, XFS is currently stable and designed for large files. ext3 is usable; ext4 is a bit bleeding edge and people have lost data using it.
• If you can't handle the volume on the mythtv users mailing list, there's an indexed archive that you can search.
• The speaker repeated the common wisdom that an NVidia graphics card with the NVidia binary-only driver "just works". This has been very much not my experience, but it must be true for lots of people.
• He admitted that his first install of MythTV took a week of hard work before it "mostly worked". He says that once you have the setup working, using it (even for the non-technical) is no harder than a commercial system (such as Tivo), and administering it is a couple of minutes a week for an experienced Linux user, but setting up is definitely harder than it should be.

I just upgraded my computer hardware, so I'm typing this on a shiny new computer with 4 cpu's, a terrabyte of hard drive, and 8 Gigs of memory.

Getting all the stuff from the old computer to this one is still harder than it should be, but is easier when you upgrade while the old system is still working.

What ought to be true is that if you move the /home directory, install the same set of packages, and import the data from the database, you should have a working system.

Some people claim that you can just copy /etc and /var to the new computer and then the new system will work the same way the old one did. I didn't find this to be true, and I've been hand moving the things from /var and /etc that I turn out to need, or reconfiguring the new system. Part of why this is less true for me is that the new system is a 64 bid install, and the old one was still 32 bits.

In any case, when you have the luxury of the old system still working is a good time to check that your backup procedure is working, and to add things to it as you find yourself manually moving something from the old system to the new one.

The most embarrassing hole in my procedure was that I found my system for entering lilypond into emacs via USB keyboard didn't work because I'd installed a little program it needed in /usr/bin (which should be only executables from the package management system, and doesn't get backed up) instead of /usr/local/bin.

I don't yet have either gallery2 or wordpress working right on the new system, but the old system seems to have the same problems, so it probably isn't the backup procedure.

My own backup procedure is largely rsnapshot, along with some scripts that back up databases and the websites that are hosted elsewhere. This gets everything you need (as long as you tell it the right files to back up), but is fairly large and cumbersome, so one of the things I'm missing is recent off-site backups. It backs up to a 1 terrabyte firewire drive. Each backup takes up about 160 Gigabytes, but the files that are the same are hard linked, so 10 backups are only about 200 Gigabytes.

Anyway, I'm very happy with the new system, because now when the backup procedure starts I just barely notice, instead of having to stop what I was doing on the computer and go get a cup of coffee.

Until the recent urge to upgrade my computing environment, I was using firefox as my main browser. This was not because it's a particularly good program, but because it has managed to get enough market share away from Internet Explorer that the people who design websites test on it. There are very few websites, except for the ones that deliberately try to restrict themselves to commercial OS's, that don't work on Firefox for Linux.

Unfortunately, there are at least two major problems with Firefox on Linux, that cause me to try other things from time to time:

• It has major memory leaks. This means that if you keep your system running for weeks, or even days, at a time, after a while Firefox will grab enough of the memory that everything, including the program that draws the windows on the screen, is gasping for memory, and going to read the hard drive before drawing every pixel. I knew that killing firefox would fix this, but I kept not getting around to it until killing firefox took several minutes, because of waiting for the pixels to redraw.
• Less common of a problem for me, but still a sign of an uncivilized program, is that once you play music with firefox, it grabs some sound resource, and doesn't let it go, so the next time you want to run some of the other music programs I use, you have to kill firefox to do it. (For those who wonder, the way to do this is the command "alsa force-reload". This kills all the programs that are holding onto resources that prevent alsa from reloading the modules it uses.)

Now some people claim that opera is a good browser for their purposes. Whenever I download it and try it, it takes me about 10 minutes before I find a site that doesn't work with it. I believe the site I do my online banking at is a frequent offender, so if you have a bank that tests their online banking with opera, maybe you don't have my problems. But switching banks is even harder than switching browsers, and by most of the bank evaluation criteria I use, mine is pretty good, so I'm not going to drop them just because their software testing leaves something to be desired.

A lot of the other browsers for Linux are in fact using the mozilla engine, which is the same one Firefox uses. Of course, I don't know exactly where in the code these memory leaks and resource hoardings happen, but it wouldn't surprise me if a different browser using the same engine had the same problems.

However, it is possible to install an alpha test version of google's chromium-browser for Linux. The version for windows has been out for a few months and gotten rave reviews for being clean and fast. The Linux version has a lot of things that don't work, but for the ones that do, it's really a lot of fun to use.

It's a nuisance to be doing testing on alpha software (which was pre-alpha until a few days ago). And of course I need to restart chromium-browser every day when I get the new copy. And do some testing to find out which things work today. Yesterday was very exciting, because "copy link address" worked for the first time. But today, it doesn't seem to want to display some slightly complicated PHP for my wordpress administration, which it's been doing for some time. Another button that's inconsistent is the "Publish" button when I publish a new post. And of course, they aren't even claiming that printing or flash work. I also still haven't reconfigured emacs to use chromium instead of firefox, so when I click on a URL in my email, I still get a tab in Firefox.

So I do still have to keep a copy of Firefox running, but it usually only has one or two tabs on it, so it isn't that much of a nuisance if I have to restart it, and it actually behaves very much like a civilized program if you're closing most tabs right after you open them.