KDE is awesome largely because KWin is awesome. If I had to name one reason I gave up using Gnome and moved to KDE, it'd be KWin.

Some programs (for example, Songbird, aTunes, Google Chrom(e|ium)) try to manage their own windows in silly ways like hiding their title bars and window borders, or fiddling with how they appear in pagers or task managers / application lists, or trying to manage their own sizes and positions. KWin lets you override and force sane preferences upon such programs.

Chromium for Linux is nice enough to give you an option to "Use system title bar and borders" if you right-click the top. But this only changes how Chrome looks; it doesn't make the KWin title bar and borders appear.

Why would you want KWin title bar and borders? Because by default, Chromium's self-managed title bar only gives you Windowsy options like minimize/maximize/close. It doesn't give you Linuxy options like double-clicking the title bar to window-shade the window, or right-clicking the title bar to get the KDE menu to send it between desktops or make it Always on Top, and such goodness. If I wanted gimped up window management I'd go use Windows 95.

A month or two ago I wrote a little script that starts saving HTTP_REFERER when people visit my site, looking for search strings from Google, Yahoo, and MSN. This is similar to one of the results Webalizer gives, but Webalizer seems to seriously undercount searches for my site. I wanted to get a semi-accurate idea of what people are searching for and how people are finding this site. The results are certainly interesting, to me anyways.

• Google sends an order of magnitude more traffic to me than the other search engines. MSN is next in line, but MSN seems to be double- or triple-counted by my script. I don't know what the heck MSN is doing, but it registers multiple hits right in a row every time anyone clicks a link from MSN to my site.
• So far as WHAT people are searching for, most people are searching for origami, which surprises me since it's not something I update very often here and I almost never post in my blog about it. The next most popular search is for conky. Then mostly problems with KDE and compiz/beryl; looks like lots of people have problems with those.
• There are also some fairly nasty things showing up in my logs. I won't post them because I don't want Google to spider those search terms and re-enforce the results. But yeah. Think about the internet, and use your imagination. (And keep your imagination to yourself. Google reads my comments too, I'm sure.)
• It's really amazing how quickly changes are spidered and integrated into the search engines. I'll post about something and just a day or two later people will start showing up here after having searched for it at Google.
• In the olden days you generally had to submit your site to a search engine so that it would know your site exists, and then pray that it decided to spider you. I remember a years-long struggle to get one of my other sites to appear on Yahoo, starting back around 1999-2000. Nowadays it seems like search engines are much more adept at "finding" your site, and perhaps much less picky? I have never taken even the slightest action to get my site to show up on Yahoo or MSN, but there it is; and this site is arguably of little or no interest to the vast majority of people in the world.
• There are plenty of people searching for my first and last name, which is kind of scary in a way. Sometimes it's easy to forget that that the things you're writing on a web page are going to be visible to a large number of people all over the world, and probably are going to be cached for a very long time and associated with your name. However it's possible people are searching for other people with the same name as myself; there is a US judge with the same name as me for example (but my site is higher in the search results than things relating to him; that's probably a sad fact, actually).

As 73 other blogs already mentioned, you don't need XGL to run Beryl any longer if you use the beta nvidia-drivers in Portage (which are currently quite masked). This is pretty nice because the last version of XGL that worked for me was from July. I've had to mask every one since then. I got some kind of GlxBadDrawableSomethingOrOther error every time I tried to run the new ones. I tried everything I know of (which is arguably not much) to get the newer versions to work, but nothing ever worked right. Tried every version of nvidia-drivers under the moon, recompiled my whole system once (for other reasons, but still). Nope. So this is good news. It runs at full speed too, which is also nice. A few short weeks ago, nvidia drivers + Xorg would crawl any time any special effects happened. Once again I'm amazed at how quickly the development of this project is going.

I noticed a lot of themes missing from Emerald that used to exist in Compiz/CGWD. Apparently they removed all the non-GPL themes, which is good I think. We don't want Beryl being sued out of existence by MS and/or Apple. However I don't think most of the themes there now really take advantage of the flashy transparent eye-candy that Emerald is capable of. If only I wasn't working on 12 other things and dead tired from work, I might give a theme a shot.

I noticed recently some (new?) feature of Google where it stores your search history and then personalizes your search results somehow based upon your stored history. When using this, Google by default stores all search results any time you do a search while logged in. You can also browse through your history and see all your past searches. Apparently this "feature" was automatically enabled on my account, because I sure don't remember enabling this in my Google account at any time.

This either crosses the line or is really close to doing so, in my book. I don't want some corporation storing my search results. It's worse than someone looking over my shoulder while I'm using the computer (which I also hate); it's like a couple thousand people I don't know, staring over my shoulder, taking notes. I don't know what Google is doing with all that data. Well, strike that, I can take a good guess: making money from it, somehow, probably involving selling the information to ad companies or using it internally to target ads at people. The concept of "targeted advertising" makes me sick to my stomach. It's bad enough being manipulated by people trying to squeeze money out of me. But being manipulated via an automated human-manipulating computer program is a totally abhorrent thought to me. It's an unethical use of computers in my opinion. Especially in any case where it's thrown at me without my knowledge or explicit consent.

You can disable (Google calls it "pause") the stored search "feature", and you can remove items from your search history, or clear the whole search history. But again, who knows what's going on there. Is it really being cleared? Is it just being hidden from your view? Quoth this page:

What happens when I pause the service, remove items, or delete the Personalized Search service? You can choose to stop storing your searches in Personalized Search either temporarily or permanently, or remove items, as described in Personalized Search Help. If you remove items, they will be removed from the service and will not be used to improve your search results. As is common practice in the industry, Google also maintains a separate logs system for auditing purposes and to help us improve the quality of our services for users. For example, we use this information to audit our ads systems, understand which features are most popular to users, improve the quality of our search results, and help us combat vulnerabilities such as denial of service attacks.

Emphasis mine. "Removed from the service"? Not "Removed from the servers", eh? Sounds like a bunch of weasel words if I ever heard any.

This crap should have been disabled by default, not enabled. If it should even exist in the first place. This makes me really happy I stopped using Gmail. I basically avoid logging in to Google at all times, at this point. And I have Firefox set to delete cookies every time I close my browser.

I've blabbered on quite a bit about finding a good webmail client since I ditched Gmail. I am enjoying being free of ads and knowing my email is not being stored / used / read by some corporation. Well, except potentially by the company that hosts my website. And by my ISP. And maybe by the government. But other than that I'm good.

I think I've finally settled on Horde Imp. I guess this is one of the (few) times I've decided to go with "tested and stable" rather than "new and shiny". It's often so rare to even have the option of "tested and stable" in the Open Source world, I may as well take advantage. Given that I almost missed a job offer due to an email ending up in the wrong place, I think email is a good candidate for sticking with something that works.

A perhaps valid question that comes to mind is why do I even need webmail? There are other options. In the days of my Linux youth, I used fetchmail and procmail on my local machine to auto-fetch my mail every 5 minutes, and store it all locally using courier-imap to access it. There are some advantages to that: I could auto-sort the mail into local IMAP folders. I could fetch from multiple remote email accounts. I could open my mail locally without having to download anything; it was all already download.

However there are also distinct disadvantages. Namely, that my computer is a desktop machine. I'm pretty good about making backups, but even I forget sometimes. On the other hand a remote host likely has someone paid full-time to do backups on a regular basis, and there is likely a huge incentive to actually do it consistently (i.e. if they fail, they will lose customers). Also, given that all sorting is done only once it hits my local machine, accessing mail MUST be done through that local machine. So if I want to view my mail remotely, I have to remotely access my home machine. This is a problem because, again, my local machine is a desktop computer. I don't have a backup power supply. I have a cat who likes to head-butt my power button. So there are times when I'll be stuck somewhere without access to my email. Again, uptime is something I think my website host is better at than I am.

Both of these issues can be semi-solved by leaving a copy of all mail on the remote hosts, and only fetching a copy to my local machine; then I can fall back to using the remote hosts in case my home machine is ever down. However this is a hack at best. Accessing a server full of 5,000 unread unsorted emails isn't my idea of fun. Far better to use the remote host directly all the time. Then I don't have to worry about backups (as much) and access is exactly the same from home or from work or from anywhere else.

Another option is to use a desktop-installed email client program. The only problem with that is that I'd have to install a client it everywhere I go. I am away from home far more often than I am at home, nowadays. Home is where I sleep while I'm waiting to go back to work. Given this, webmail is about the only option I see open to me. A lot of computing tasks really becoming more distributed nowadays, aren't they? I couldn't manage if home was the only place from which I could access my email. Interesting.

After two years of using Gmail, I have exactly 5,531 emails. These are all sitting on some server who knows where, and getting them all off of there and onto my own server was an interesting task.

It turned out to be pretty easy thanks to Thunderbird. If you enable POP in Gmail, you can use Thunderbird to access Gmail that way. So I set that up, and set Gmail to serve up all messages including read / archived ones. Then I let Thunderbird fetch them all. This took a while (maybe an hour?).

The thing that makes this work is that Thunderbird stores mail locally as plain old mbox format files. So if you browse to ~/.thunderbird/blargaljdf.default/Mail/pop.gmail.com, you can see files with names matching the "mailboxes" in Thunderbird. There are multiple files per "mailbox". The one without an extension is the mbox file.

All I needed to do was upload these mbox files to my IMAP server and subscribe to them, and that's it. I actually used Thunderbird to split up all those Gmails into multiple mbox files first and then uploaded those. The Thunderbird GUI makes this all fairly easy. I have no use for Thunderbird normally, but it's good for this task.

It strikes me that this would also be a good way to make a backup of Gmail, if someone still plans to use it. Or any other POP email service, I suppose.

Interestingly, doing this caused IlohaMail to meltdown. It would say it knew I had 500 emails in folder X, but it would only let me see 120 of them. So now I'm using Horde again. Someday, I'll find a webmail program that works. Someday.

My host comes with Squirrelmail and Horde, and I dislike them both. Squirrelmail is too bare-bones for me. It's hideous to look at. And Horde is too over-the-top and slow for me. It takes about 17 clicks to get to the inbox once you log in. No call for that.

Really, what's the difference between one webmail app and the next, except the interface? So I'm looking for the nicest interface I can find, so long as the backend still works too.

So first I tried RoundCube, the much-acclaimed AJAX webmail app. It's pretty impressive in the "Wow, someone got a web browser to do that" sense. You can drag-and-drop things etc. But you can tell it's an AJAX hack. Things subtly don't work. If something is loading in the background, and you try another action, things often freeze or break. The folder list on the left doesn't update correctly; you have to click off a folder and click back on it again to get it to reset the message count. If you move some items to a folder, and quickly move some more to a different folder, one or the other operation will silently fail. Etc. etc. The problem (with all AJAX apps) is that there's all kinds of crazy server communication going on in the background and you never know what it's doing or if it's done or what. At least in a traditional web page you click a link or button and then you see it loading and you wait for it to finish. And then you click another.

Other things are broken in Roundcube too, like putting a < in a subject causing the subject to be blank. Probably some kind of relic of having an interface that is itself made of a bunch of changing HTML tags. I started the bugzilla for RoundCube and there's just an enormous number of problems with text parsing and encoding etc. So yeah, I passed on this. Maybe in 6 months if it's more stable I'll try again.

I considered Hastymail next. But development has apparently been stopped for a long time, and although it's not as bare as Squirrelmail, it's possibly a bit over the top in terms of "Let's make an interface that looks like it came from 1992". I like minimalistic, but Hastymail looks huge and clunky to me. I don't think it has threaded messages either, which is one thing I really love(d) about Gmail. Hastymail does have some neat extra stuff like showing emails from multiple folders at once in a multi-levelled list, but none of it looks all that useful.

I finally settled on Ilohamail. It's fairly minimalistic, but it looks nice. And it has just the slightest touch of AJAX, like threading conversations into expandable blocks in the message list. Development also seems to have stalled on it for a while, but we'll see how this goes.

And now, in any case, my emails are no longer being auto-read, stored away and turned into ads by Gmail's server farm. Sweet freedom.

Email is such a fragile thing. It's nerve-wracking setting up an email server properly. If you set it up wrong, emails disappear into a black hole and it's sometimes not readily obvious that there's even a problem, especially if the emails don't even bounce. Same goes with setting up a procmail filter, for example.

One of my old CS professors said there are three kinds of programming bugs:

1. Bugs that break things in obvious and catastrophic ways, causing a complete crash of the program.
2. Bugs that break things without causing a full crash; the program keeps running. But you know about it.
3. Bugs that break things without causing a full crash; the program keeps running. And you don't know about it.

In that order, they increase from bad to worse. By far the worst is when things look like they work, but they don't. You can't even begin to fix a problem until you know a problem exists. (Programs' sweet debug-info-filled death-cries are programmers' best friends.)

Emails vanishing without being delivered or bounced is the third kind of bug, i.e. really really bad. My host's email system is very hard to work with and leads to many of this kind of bug; mysterious vanishing emails. All I want is all emails to both of my parked domain names to go to a single address at a single one of my domains. I'm pretty sure I got it right this time, but I always think that and then two weeks later someone calls me and says "Hey, how come you never responded to my email?" and it turns out the server ate it for dinner. But I will admit, it's very likely my troubles are PEBKAC-related.

If I get this set up, I'm going to stop using Gmail once and for all. I'm tired of 1) being advertised at while I read emails, 2) the Gmail servers being either down, non-responsive, or dead-set on not giving me my emails, 3) having stupid extra options to wade through like Google Talk and Google Calendars and "Invite your friends to use Gmail! We need more ad-reading monkeys!". And consarnit, I miss real folders. Labels don't do it for me.

Email providers are really in a good position. Once you start using an email address, it's REALLY hard to stop using it. So once people are hooked into using your service, you have them in your claws and you can start milking them for all they're worth. Imagine how many people, organizations, mailing lists etc. out there have your current email address and how long it would take to get them all to switch over. The only good thing is if you can forward from your old address to your new, and slowly do the transition that way.

So today at work, Gmail suddenly died. I couldn't log on to the Gmail webpage, and it was giving a "File Not Found" error instead of an actual Gmail error page. I tried logging on to Google in every way I could think of, and nothing worked. I tried logging on to Adsense, and the page loaded OK, but on the right instead of username/password fields I got the same File Not Found page.

I tried another computer at work, and it worked fine. And now I'm at home, and it works fine here too. I tried deleting my cache/cookies at work, and I tried it in IE and Firefox, and nothing worked, so I don't think it's a browser problem at work. Maybe something screwed with my hosts file? Maybe Google is blocking me via IP address?

The sucky thing about email addresses is once you get one and give it to tons of people, you're stuck using it forever. I really don't like Gmail at all, but it's the best I can find. I tried Horde and/or Squirrelmail at my domain, but it's just not powerful enough and I don't have full access to my server to actually configure those things like I want. Someone needs to make a simplistic, powerful, geek-based email service. (Or is there such a thing already?) I don't need fancy options. I don't need AJAX. I don't want emoticons or stationary or God only knows what else. I'd use anything which was reliable and fit those criteria.

I just had a nasty run-in with PAR. Activestate Perl doesn't play as nicely with CPAN as I'd like, so I usually use its own built-in "Perl Package Manager". The problem with that is that installed packages aren't rebuilt (from what I can tell), they're just downloaded in binary form and unpacked. So you often don't get the latest version that's on CPAN.

PAR was choking on Scalar::Util, because it couldn't find a method named "refaddr". But the program worked fine when run as a .pl; it only choked after being pp'ed into a .exe via PAR. I only need to make a TINY change to this script and re-compile it into a .EXE, so I wasn't really looking forward to rewriting this Perl application just so PAR could read it correctly.

Thankfully I found an answer on Google groups. I coulnd't tell you how many times Google groups has saved me a heck of a lot of time and/or effort. In this case it probably saved me a day worth of work. Even on such an obscure question as this.

Now that I think of it, some advice: Don't use PAR for applications that are going to be used by anyone, or relied upon by anyone. It's not a lot of fun.

Actually, some better advice: Don't use Windows for applications that are going to be used by anyone, or relied upon by anyone. That's better.