Older blog entries for slef (starting at number 30)

Hacking is mostly concentrating around scheme not directly related to site-engine just yet, but should be very useful. I've decided that MVC is definitely too slow for a web site and actually factors the code quite badly in many cases. I've implemented something close to PLT Scheme's xexprxml in guile and am using it to hedge my bets. I'm keeping on with the fasttemplate, though, as it's the simplest way to use contributed templates. I'm also considering rewriting my CGI interface to look like PLT's.

The other thing I've got in my head but not yet on disk is the new database query interface. I think it should be possible to wrap everything in this, but I need to get it written out in full.

Finally, I need to package and release my IRC scripts. There are some changes I want to make from my brother's version and I want to do per-channel colouring, as I'm sat in lots of channels with lots of idlers, so I might as well have window-per-server instead of window-per-channel.

Health is apparently bad. I don't exercise as much as I used to and I'm feeling the onset of the cold this year. I need to chill out and do more walking and cycling. Went back to the surgery and was told that my blood pressure is better, but still high. I need to get this under control. With my heart and my teeth, I've just not been taking care recently. Working too little but working too much.

Work is ticking over. I need to do the hacking to give us cool new toys to play with. Hopefully we'll have a new colo box within a couple of weeks and then I can take up the slack capacity with all sorts of fun projects ;-)

I find it somewhat ironic that the term "Open Source", which was created in part to avoid the "freedom"/"cost-free" ambiguity, has developed an ambiguity all of its own with all these various "look but don't touch" source openings. If we're going to have to explain a term to clients, let's explain the older and clearer one.

I know I'm meant to grow out of this, but greve just certified me. Whoo-hoo.

I've been posting in the SourceForge drifting comments today and I must say that the responses of Pat and Chris from VA have done nothing to change my opinion of SF. They are not playing nicely with the others and I desperately hope that we get all alternative hosters and associated sites (eg software directories) involved in supporting a SNG/CoopX XML format as soon as possible. I would love to write a desktop tool to help with the preparation of software release announcements and the like which supports any new XML format for submission to supporting web sites.

Perhaps the reluctance to support it is because SF is VA's current great hope for profit, but if the rest of the world can present them with a universally-recognised format for project information interchange, they'll have to support it. They know which side their bread is buttered.

I'm also disgusted by their "secure because it's obscure" comment to hacker. How dare they try to preserve their market share by spreading FUD. I've seen some of his work and I dare say it's a lot better than the sf code...

I think I should write this at the start of the day, as I never seem to do any hacking in the first hour anyway. So, this is what I did yesterday (Monday).

I made some biiiig changes to the site-engine infrastructure. It's not big and it's not clever, but now it uses one db connection per page and will be easier to move to one per server when we get to running in an app server. The other improvement is a form input validator. Most of the appearences of the "boom screen" (my error handler) are from bad user input, so being able to give them helpful feedback on what's actually wrong is going to be a really good thing. Sadly, this combination means that I need to modify all the data and glue code (Model and Controller code in MVC-speak), but at least it should make it quite a lot simpler.

Frank (as) seems to be back on the site, which is good. He also sounds to be about ready to start Sapphire 2, so I've started throwing my requests in.

My Palm 3 is ill. I blame a hack, but it's eating batteries faster than I can charge them, so I've done a reset and I'm trying not to use it today while I charge a new set, which hurts, especially as I've got a tool to read newsgroups on it. I really must post that on my web site soon.

Other than that, it's work as normal, doing a mix of hacking and trying to track down new project opportunities. Which reminds me: I've some more email to answer...

mwh: I mean it, unless someone has added things like useful interfaces to common widget sets and access to common relational databases to Common Lisp since I last looked. Whatever language you use, you're going to need some "non-standard" libraries to do real work in them. Either that, or it's going to be in constant flux as the things it interfaces to change.

Anyway, I'm happy today. I implemented the cookie handler module I wanted and it was surprisingly easy. It's not sophisticated, but it works. Now, next challenge!

chalst: There are scsh implementations for other schemes, either existing, underway or desired. Guile definitely has scsh features and PLT schemes want them.

Is the lack of a large standard library any big problem? The libraries exist for most things and they usually interface in a similar way or you can wrap them up fairly easily. Guile also seems pretty easy to interface to C, so that's why I think I've seen more libraries using it than other schemes, although PLT offered (and maybe still offer) a guile drop-in replacement. I assume you know about slib, as that could be taken as a "standard library" for many schemes.

The other *languages* you mention are no better for doing practical programming if you just want the standardised subset of them (with the possible exception of python) -- you still have to add things to make them palatable. Lack of a standard library mechanism could be pointed to as a bigger problem, but some of the same concepts seem to be getting acceptance, namely modules with signatures, so maybe we're seeing convergence there, too? slib also has its own ideas on how to solve this, I think.

As to practical? I'm currently having some pain with missing features from scheme when trying to program in less powerful languages. *sigh*

Well, the GA Engine is working and I've set it loose on the cluster. I clearly need to do something to the cluster to let the pure mathematicians soak up idle time, but let the rest of us get on with our work. They've maxed out a couple of nodes right now, so Dr Death will have to pay a visit soon.

The web-site-engine is still waiting for SQL and more docs, but some have started to appear and I'm appealing for beta-testers and more developers to knock down this TODO list.

Well, this is why I've been a bit quiet here lately: the first website using the web-site-engine code is live, although I make no promises about its stability yet. It does appear to work, though, and using it in anger is a great way to find obscure browser bug problems. It's also been a great way to find out why I'm not a database designer -- again. Oh well.

It's only 1300 lines of code, but I need to polish the comments, get the current db schemata and write some top-level docs before I can really advise anyone else to try it. It knocks php scripting into a cocked hat for me, though.

As happy as a dead slug...

Spent a frustrating week of (non-)development on one of the websites this weekend. The grand plan remains unchanged and I think it's viable. However, one of the problems I had, I sent on to the mailing list for a project I depend on. The reply came back that I should use a beta which can only be got from undocumented cvs. I think I've pieced the bits

together (and I'll know when it finishes downloading), but I'm a bit skeptical about using this thing on a production server. Academics! ;-)

murrayc still sets mailing list filters up by hand? Primitive ;-)

Oh well... "officially" start new work tomorrow I think. I suspect someone will be glad.

Interesting. A large number of people seem to be surprised that they hit journeyer status at about the same time as I did. Maybe one of our trust metric philosophers out there could do a little investigation and write an article about how we got to this sitation. I think I know how I got here, but maybe it's all part of a sudden upsurge in certification by a group of people, or just new memberships. Also, the conventions of self-certification would be an interesting study. If anyone's got time to burn ;-)

Have to agree with the comments about more and more vapourware projects appearing on the free hosting sites. SF seem to be the worst offender. I know my project on savannah hasn't got any code available just yet, but that's partly because I decided to stop and rewrite the lot.

Important meeting for New Work deferred. Oh well.

rmorrell's recent comment on the Stallman Saga is both enlightening and troubling to me. For those who don't read newsgroups or the lists I'm on, he has been getting criticism for breaching the GPL with a beta distribution (you too can play hunt the 0.9.9 beta source) and flaming people who asked him about it, although he did have previous form with another company from the group Jason works for.

Looking back at RM's comment, there are some interesting parts because of this. I highlight particularly: "[FSF] assisted us greatly in copyright violation issues with our SmoothWall project"; "Why do managers and decision makers not want to use Linux/BSD solutions over proprietary solutions [question not answered]"; "OpenSource" [not Free Software]; "Arguments are GOOD, dirty washing in public is bad, but confrontation can be good for development teams"; and I leave the reader to draw their own conclusions.