Older blog entries for jacobsan (starting at number 3)


Wow, long time no update, eh? Guess I slipped back into putting diary-like stuff on my weblog -- hafta work on that.


I've ended up giving up on the C++ study group at work -- I wasn't really getting anything out of it, and I think I've picked up enough that (a) if I have to hack on somebody else's C++ code I won't be totally lost and (b) I'm not going to be using C++ for any of my new projects anytime soon -- I'll stick with Perl and C, thanks much.

It's still not clear to me how much of my disgust with the language is due to the language itself, and how much is due to the awful C++ Primer, but I'm pretty much beyond caring.

I'm currently making another assault on Goerzen's Linux Programming Bible. Previously, I've crapped out about eight or ten chapters in, but I'd really like to absorb more of the 'C mindset', so I'm giving it another go.


My recent upgrade to the Helix^WXimian Gnome 1.4.1 .debs broke GMC in a number of interesting ways. This was a PITA, since I'd been storing links to check out later as desktop URL files. So, I sat down, thought for about schemas for about 15 minutes, did a bit of CGI/Perl/SQL hacking, and whipped up a little bookmark mananger. I realize that this has been done tons of times before, but I've played with most of the bookmark utilities on freshmeat.net at one time or another, and found most of them unsatisfying for one reason or another. Plus, it's nice to have a little project like this, where I can hack in new features one at a time, but still keep a decent overview on where the whole thing is going, so that it doesn't end up as a huge mess (he said, crossing his fingers...)

Actually, I think I can see a way to easily extend this into the poor man's CMS/weblogging tool that I've been thinking about for a number of months. Now I just have to decide if it's worthwhile to do -- I'm not sure I'd stick with a tool where I had to do extensive text editing in a browser. OTOH, getting the backend in place with a CGI interface that talks to it is probably a good precursor to making the backend talk to XEmacs. On the gripping hand, I am gonna keep fooling with the damn thing, so I might as well try to plan ahead a bit...


It's getting to be time for me to start thinking about what I want to be when I grow up (most would say I'm a bit late...) I've got about one year left in my current post-doc. My options are to extend the post-doc (which I could easily do, for another year, or even two), try to get a Real (scientific) Job at the current workplace, try to get a Real Job somewhere in the biotech/bioinformatics industry, or try to get a job completely unrelated to science -- systems administration is the current leading contender in this category.

The Big Decision I have to make is between one of the first three and the last one -- I'm reasonably sure that any of the four alternatives are open to me, but the last one is a much, much bigger step than staying in science. There are quite a few things about sysadmin work (or what I've seen and heard about sysadmin work, anyway) that appeal to me -- I like making things work the Right Way, I like having a variety of different issues to deal with during the day, and I like 'interrupt-driven' work. The money would be as good or better than staying in science, so that's not really a factor -- it's just going to be a pretty irreversible step.

Ah well, it's certainly not something that needs to be decided overnight. Any advice?

nymia: My point (which I realize I didn't really bring out) was that, unless you define more exactly what you mean by "merit", advocating a meritocracy is basically semantically null. I gather you're in the Phillipines; you could argue that you've got a meritocracy -- it's just that the merit isn't in what you can do or have done as much as it is in who you know. (Not that different from the US, actually...)

Of course, defining what merits you're talking about is only half the battle, of course. You've still got to come up with a system that allows selection on the basis of these merits, which doesn't rapidly devolve into "whoever the testers like gets the job"...

Look at what we can do...
Bah. I've had five or six pages of object data structure and method prototypes written out (by hand, because I find I think better that way) for a couple of days now, but haven't had time to put them into the BOP wiki that my co-developer kindly set up.

Of course, it didn't help that I finally tired of the various problems I've been having with the Helixcode^WXimian Debian packages, resulting in my uninstalling all of them and trying to build Gnome from source (which I've done successfully before), only to give up in disgust as I realized I couldn't use stow on each individual package, but instead had to throw them all into the same directory. Only took two days in console-land for me to figure that out. Bah, I say.

However, since I stayed home sick from work today (just general crappy feeling -- a cold combined with lack of

sleep, which could have turned into something worse had I not spent most of the day zonked out), I was able to re-install all the Helixcode^WXimian packages via apt, and do a general tweaking of my installed packages at the same time. Downloading a couple hundred megs of debs takes a while -- which was good, as I needed to nap instead of dink on the computer. 8^)=

nymia: Sure, a meritocracy sounds good. Which set of merits are we going to go by?

I'm wondering...
Is it just me, or do mirwin's diary entries look like somebody has trained some sort of Markov-chain-based text generating bot on a couple gigs worth of old diaries and set it loose? If that's the case, I really wish the bot owner would set the output to a shorter length.

I'm taking part in a group at work that's reading through Lippman and Lajoie's C++ Primer. We're up to chapter 8, I think, and it's been completely disappointing. (I guess I should preface this by saying that while all the code I generally write is in Perl, I have messed around in C, and I've read a book or two that touched on or intro'd C++ concepts.)

First, the book is a hideous mish-mash of core C++ stuff and STL techniques. Worse, fairly advanced concepts are touched on, sometimes at length, without any background. For example, there's a vector class implementation at the end of chapter 2 (IIRC). Great -- but we don't know what a class is, OOP hasn't been mentioned at all, let alone things like function and class templates.

So, I'm getting a very negative view of C++ from all of this, one that it may or may not deserve. For instance, the whole language seems huge! The syntax feels really ugly too. The more experienced hands at work are telling me that this is just due to the mix of core and STL in the book, and that the actual language isn't that bad; it's just that the STL sucks.

Anyway, I'm sticking with it for the time being; it certainly can't get any worse. Does anybody out there have any 'intro-to-C++' books that they can say good things about?

Guess it's time I started making use of my Advogato diary. I've been reading other peoples on my Visor, using jmason's excellent sitescooper tool, and I figure it's time to start contributing as well.

Haven't been doing all that much coding lately, at home or at work -- just too busy with other stuff. I did emit a bit of code last night, just a little Perl to drive analog and ncftp. Once I get done, I'll have a tool that let's me automate fetching and processing log files from my weblog. So far, I've got it fetching the log file for a given day, and the completed log file for the prior day, and running analog (with slightly different configs) on each. Next up is adding in the logic to produce rolling weekly and rolling monthly reports, which is going to involve cat-ing together the appropriate log files.

Eventually, I'd like to be able to auto-archive the log files by month, but I'm still not completely sure how I'm going to do that without it being a huge kludge. In fact, I'm not sure I'll ever be able to generalize it enough to make it useful for other people, but we'll see, I guess.

Right now, however, it's time to write a weblog update, and then catch-up with NYPD Blue on the Tivo...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!