Charl P. Botha's Journal
[Most Recent Entries]
[Calendar View]
[Friends]
Below are the 20 most recent journal entries recorded in
Charl P. Botha's LiveJournal:
[ << Previous 20 ]
| Friday, February 7th, 2003 | | 1:39 pm |
| | Wednesday, January 29th, 2003 | | 2:28 pm |
Valuable lessons learned today...
After having gone back to my workstation vendor for the 4th time in so many months today with a broken graphics card and after having had to replace the now scarce SMP motherboard, I've learned two very valuable lessons:
- When you buy a workstation for Real Work(tm), get at least a three year on-site guarantee. You might think that you can't afford the extra EUR150 (or roundabout), but after you tally up all the time you might have to spend fixing your own workstation during its lifetime, you'll realise that you're wrong.
- When given the credit card at the beginning of a four year project to buy the machine of your dreams, DON'T splurge. Buy a somewhat more modest machine, but make the arrangement that you'll have the right to purchase a brand new computer half way through the project. Remember how fast computing equipment ages.
| | Sunday, January 26th, 2003 | | 9:55 pm |
| | Tuesday, January 21st, 2003 | | 2:03 pm |
I reveal myself
My Meyers Briggs personality type is ENTJ. I don't know whether one should pay any attention to these kinds of things. Personally, I believe it's mostly bullshit. :) | | Monday, December 23rd, 2002 | | 10:44 am |
| | Sunday, December 22nd, 2002 | | 5:31 pm |
Porting woes This is brilliant. Microsoft Visual C++ violates the ISO C++ standard (at least as far as my 1998 document is concerned) with regards to this very fundamental and simple scoping behaviour, but they do not admit that this is a violation as such. Instead they claim that one can work around this "problem" by using a special compiler switch that disables "language extensions". However, depending on your configuration, making use of this switch will break compilation of certain Visual C++ headers (ntheader.h IIRC). You have to love this. | | Wednesday, December 4th, 2002 | | 11:17 am |
| | Friday, November 29th, 2002 | | 11:51 am |
AAARRGGHHHH!
Another one of my favourite online radio stations shuts down thanks to CARP. Evil Dildo, RIP. Current Mood: pissed off. | | Tuesday, November 26th, 2002 | | 10:39 am |
Yoohoo!
I'm back from an incredible holiday in good old ZA. Now I just have to survive my article deadline on Friday and then everything will be just dandy. Fortunately there's RantRadio to keep my industrial ears happy. | | Wednesday, October 23rd, 2002 | | 12:00 pm |
The USA loves OIL!
This is brilliant: "I helped blow up a Bali nightclub -- by driving my SUV to work every day!" Thanks for the link, Rudolph. Read the whole article here. Keep in mind that the USA is responsible for a quarter (25%) of the world's TOTAL oil usage. This and other interesting tidbits such as the fact that the USA also burns 25% of the world's TOTAL coal are available here. | | Monday, October 21st, 2002 | | 3:59 pm |
Patents suck, HARD.
Damn it, when will these idiots learn that patents are evil?! This article (which might be Slashdotted at the moment) reports how a certain genetic test for breast cancer can not be performed by a British Columbia hospital anymore because the morons of Myriad Genetics have patented the two genes that can signal whether a woman may develop hereditary breast cancer and have legally threatened the hospital. How the F*CK can one patent a gene? Nevermind the very apparent stupidity and lack of reason behind this (the gene was already there, the company merely discovered it - there's NO creative innovation here!): just practically speaking, this is screwing the populace over. This test now costs three times as much, just because Myriad went to the trouble of taking out the patent at the even more idiotic patent office. ARGH! | | Monday, October 14th, 2002 | | 4:20 pm |
Urgh, mail server meltdown...
A DIMM in one of our mail servers decided to break today... to make a long story short, I had to install a new imap on my own machine (as I switched incoming MXes) and with the new installation a new SSL key was created. Evolution (the MUA) of course kept on trying the old server key. DOH. After quite some minutes of swearing, re-installing and doing various other pointless things, I deleted the very suspicious-looking "cert7.db" and "key3.db" files in ~/evolution. Problem solved. This should be somewhere in the Evolution UI or at the very least help system. | | Sunday, October 13th, 2002 | | 12:24 am |
A useful tip
So, we all agree that Python is The Best Thing Since Sliced Bread(tm), right? To add several levels of pleasure to your interactive Python experience (ribbed or studded), please try IPython. This is so flexible that one could even consider using it as full-time shell, instead of your bash, (t)csh, zsh or whatever. Hmmm.... | | Tuesday, October 1st, 2002 | | 1:42 pm |
| | Tuesday, September 24th, 2002 | | 2:07 pm |
Christmas Present
If you want to buy me something, get me a laptop with one of these. Thanks, I appreciate it. | | Monday, September 23rd, 2002 | | 6:26 pm |
Good UI != Graphics?
I tried out Ximian Evolution today. It's an email client, actually more a personal information manager, that looks a lot like Outlook, except that Ximian is free, runs on *ix and will probably not get your computer infected with 298374987 different interesting yet unfriendly viruses. Hey, interestingly, it seems the correct plural is "viruses" and NOT "viri". If you're a pedant, dodge this. In anycase, usually I use mutt which, unlike Evolution, is exclusively text-based and 100% hot-key (shortcut) driven. For example, to compose a mail, I would go "m", type the address, the subject and the body of the mail, then ctrl-x-s-x-c and finally "y". The mail is sent in record-time. With Evolution, I seem to be doing a considerable amount of clicking with my mouse, which definitely slows one down. For instance, with mutt I can type "c" and it will show me the first mail folder with unread mail (all my mail is of course sorted with procmail); simply typing enter takes me directly to the unread message itself. With Evolution, I have to click on some drop down list, then click on a folder with unread mail (at least it's rendered in bold), then click on the message listing pane, then type "n" to go to the unread message. Maybe I'm just not going about this the right way? Whatever the case may be, at the moment MUTT with its far uglier text-based appearance has a MUCH better user-interface than Evolution. My interest in the Ximian product lies in its calendar and to-do lists however. I hear that syncing with my Palm Pilot is implemented really well. It would have been nice however if the mail client was somewhat more hotkey driven as this really impacts greatly on its usability. I hope 1. that there are some more hidden hotkeys I don't know about OR 2. that this gets added in the next release. BTW, I had to set up uw-imapd (so all my folders can remain in native mbox format; Evolution does support mbox, but in a very limited fashion) but my inbox is in a non-standard location. With debian, you can drop something similar to this patch into the source package's patches directory before rebuilding it to do the non-standard inbox location thing. uw-imapd is NOT very hot on configuration. | | Monday, September 16th, 2002 | | 3:32 pm |
Pragmatism
The man who is not a socialist at twenty has no heart, but if he is still a socialist at forty he has no head. -- Aristide Briand (1862 - 1932) | | Friday, September 13th, 2002 | | 11:42 am |
All operating systems suck
At least, all operating systems that I've worked with suck in some way or another. These are not minor quibbles, but major problems. It seems these fundamental problems are present in both open and closed source OSen. You've probably read The Cathedral and the Bazaar by Eric Raymond, so you know the difference between the two methodologies. Most of the open source development that I've come into contact with adheres to the Bazaar philosophy. The problem with the Bazaar philosophy is that almost anyone is allowed to contribute to the code (of Linux for instance). A few of these contributers are very clever people, but in a maverick kind of way. Whatever the case may be, they do not work together in the close-knit coordinated groups that you find in good software houses with effective leadership, but rather in a hackish manner. This wouldn't have been such a problem, were it not for the fact that the rest (and the majority) of the contributers are not sufficiently experienced, inept or just plain stupid. Partly due to these phenomena, open source software seems to suffer from a total lack of consistency (in form, function and quality). In addition, one finds hundreds of half-assed software efforts on the web that will NEVER be completed. Compare this image to what you read here for instance. Another problem is that everyone contributing to an open-source project can do more or less what they find amusing. This is the reason why documentation is always out-of-date or plain non-existent. Look at the brilliant API documentation of GTK 1.2.x for instance. Formal code testing is often also ignored. There are exceptions to this rule, such as VTK for instance. VTK makes use of a rather impressive unit testing framework and has excellent documentation. However, VTK is backed by a motivated company consisting of really clever people that coordinate the whole effort. Hmmm, I started with OSen that suck, but the issues above apply to all OSS efforts, including operating systems. One other problem with OSS operating systems is of course vendor support. When a new video card is released, the first thing the vendor does is to create working drivers for Winblows. In a few cases, vendors have created very good closed source drivers ( NVidia) or have made the specifications available ( ATI) in order to address the non-Windows community. In the former case, however, the drivers have significant problems (the nvidia drivers do not support suspend/resume or power management on laptops, the drivers are exclusive to Linux/X86) and in the latter case, the open source XFree86 driver development effort doesn't have the manpower or time to create rock-solid drivers in a reasonable time and parts of the specifications (e.g. HyperZ) can't be released, meaning that the Linux drivers will ALWAYS be outperformed by their Winblows counterparts. So, you decide to stop wasting time on open source products and buy a copy of M$'s latest and greatest: Winblows XP. Teams of very capable programmers (by all reports) have spent thousands of man-hours on creating this system. However, you now have to deal with a whole new can of worms. The OS might have driver support for every device you can think of and support all your state-of-the-art PC's most interesting and obscure features, but you have NO idea what this OS is doing behind your back and you are definitely at its mercy. M$ may download "updates" onto you PC automatically; there may be security problems that are deliberately kept from you; things like Palladium make sure that we live in interesting times; if there is some problem with component or driver, you're at the mercy of the time-schedule and motivation of the vendor, even if you do have the skills to fix it yourself. This is not such a pretty picture either. We could go down the list of OSen that are available for the PC and for real UNIX workstations and we could make lists of similar issues. There can be no real 100% reasonable conclusion, but armed with these examples (and many more which I'm too lazy to write down) I can bitch: All operating systems SUCK. | | Wednesday, September 11th, 2002 | | 11:04 am |
Happy me
As I was downloading OpenOffice for the wif last night, I couldn't help noticing that it was coming down at a very respectable 110KByte/s via my home ADSL link. This of course means that my bandwidth upgrade is complete. Rejoice! | | Sunday, September 8th, 2002 | | 11:53 pm |
Improve that UI This guide to user interface differences between Winblows and Mac OS X (by Apple) should be consulted when you ever have to think about getting that UI perfect. Apple makes several very valid and straightforward points. |
[ << Previous 20 ]
|