Monthly Archives: June 2007

You know it’s summer when …

  • It’s broad daylight at this hour (21:15)
  • You have freshly-made gooseberry fool, cherry pie, and a punnet of strawberries in the fridge, and the pods of fresh peas on the drainer.
  • Light white wines that in other seasons seem a bit thin taste wonderful, and hefeweizen becomes the best of all beers.
  • The idiot drivers are out in numbers on the roads.
  • You go for a swim at Lopwell Dam, and find not only lots of people having a wholesome good time, but some thoroughly unpleasant types in a 4×4 camping by the “no camping” signs and spreading their crap everywhere.

There are fantastic colours everywhere.  But the insects seem far fewer than they should be in this season.  Hmmm ….

Low-energy computing

The Beeb reports that the powers-that-be are giving some thought to lower-power-consumption computing kit. And not before time!

This is not entirely new. In the late 1980s I had an Acorn “archimedes” computer. The ARM processor had the computing power of an Intel i386[1], but (famously) power consumption low enough to run off the waste heat of a 386. When used without an internal hard disc, Acorn’s machines were also fanless and silent! Unfortunately I haven’t been able to find anything comparable in recent years. I don’t know if the ARM still has such good power consumption, but noone markets it in a desktop or laptop computer, or components thereof.

More recently I tried an EPIA-based desktop box: the greenest (in terms of power consumption) I could find. It doesn’t bother me that these are nowhere near as powerful as normal (Intel or AMD-based) desktop boxes. But it did bother me that the hardware was badly flaky in other ways, and has now given up the ghost. So alas I’m back to a normal-power-consumption AMD Sempron-based desktop box.

(Of course, not all is bad. Today’s monitors are a huge improvement on their CRT-based predecessors).

While the news is welcome, what we really need is the market to start to take power consumption seriously. Hitherto, with electricity being ridiculously cheap for what it is, market forces have worked in the opposite direction, as witness the demise of Acorn/ARM in the desktop market. A government initiative won’t change the profligate habits of the majority. But by raising awareness, it may at least prompt the industry to re-introduce viable options for those of us who want energy-efficient computing.

[1] At the time, a 386 was a much more expensive machine, and the ARM in fact significantly outperformed it in most benchmarks, though the Intel was much faster in floating-point.

WIFI Telephony and business models

I’ve been looking for years for a telephone that’ll combine normal GSM mobile phone with WIFI, preferably SIP. Alas, the buggers remain elusive, so my mobile VOIP use has been limited to the laptop.

I just gave up waiting for the right thing, and bought myself a WIFI phone for portable VOIP. And sad to say, it’s not SIP but Skype. I’m sure it’s a big-name/volume thing, but there seems to be a big price premium for a SIP phone over skype (for example, at Dabs it’s over £200 for SIP vs £90 for skype). So it’s sheep-like Skype for the time being, except of course when I’m at the computer.

I wonder if this is also a triumph of Skype’s business model? Skype is monolithic, well-promoted, very easy to sign up to, and largely consumer-oriented. SIP is democratic and backed by a wide range of competing providers, none of which has the profile and clout of a Skype. And the SIP providers are much more geared to providing higher-value solutions to business rather than the consumer/soho market. And the mere fact of being democratic – and consequent lack of a single signup point – is probably holding it back in the mass market.

So who should be taking on Skype in the mass-market? ISPs might do it, but (as with email) they’ll have to make it accessible from anywhere, and they’ve got a chicken-and-egg situation with the cost of equipment (could a big ISP badge a range of handsets and sell ’em cheap to subscribers?) A Big Name – a google or yahoo – could perhaps buy up a VOIP provider (or several), brand it, market to consumers and promise volume sales to the equipment manufacturers.

In due course, I expect we’ll see multi-protocol handsets (including GSM too). But how long do we have to wait?

I hate voting

Speaking as a chronically indecisive person, I hate voting!

Mostly I hate it when there are a lot of candidates I respect, and who I respect for widely differing reasons.  It’s easier when they’re a bunch of corrupt politican scumbags, and you have a clear least bad candidate or “none of the above”.  But when you want to move every candidate higher up the ballot, it’s painful.

Voting has just closed in the ASF‘s annual election of a board of directors, and new member elections.  The most difficult part was the ballot for the nine places on the board: with more impressive candidates than the nine available places, how do I pick criteria for ranking them?

  • The overall good of the ASF?  Well, who defines that?  Boils down to a mixture of their views, commitment, competence, etc.
  • Ideas and visions.  It’s not the board’s role to take on new projects; rather they support the members!  But they are overseeing rapid growth, in a pioneering environment.  That will call for both leadership and managerial competence.
  • Alignment with my own views.  Do they represent me?  Two or three candidates have expressed views I firmly disagree with, but I still respect them.  A lot.
  • Affiliation.  I’d prefer not to see the board too much dominated by employees of any one company.  That’s a matter of perception: there’s no suggestion they’d let it become a conflict of interest affecting their actions as Apache directors.  This year that is an issue.  But the company in question employs lots of people I don’t want to vote against!
  • Appearances.  Apart from the affiliation issue, does it gain or lose us anything to have our biggest names and/or more controversial people on the board?
  • What do I feel like on the day?  To an indecisive person, this is a shamefully important criterion.

One thing seems certain: whoever the nine elected board members are, they’re nine good people who will deserve and get my support.

Bizarre

This morning I had an unexpected ring on the doorbell[1].  Looking out of the window by my desk[2], I saw a policeman.  He asked for access to the back garden, explaining that some neighbour’s telephone had been thrown over the fence into the garden, and could he retrieve it.

Huh?  How on Earth is this a police matter?  Why can’t the ‘phone’s owner just come and ask for it?  Don’t the police have better things to do with their time?

Anyway, it seemed harmless enough, so I let him come through to the garden (which is shared between the four flats in this house).  He rang the number, and indeed we heard the phone ringing.  It was clearly not in our garden, but next door!

Huh?

Well, if the police have time for that kind of trivia, will they please take a few minutes each day to remove the cars that routinely (and totally illegally) park to block the pavement outside, and make the road a complete no-go area for anyone not quite agile enough to walk around them in the middle of the road (like my recent neighbour Mrs B., whose arthritis left her walking very slowly and carefully on two sticks).

[1] Actually two, but the other was just the man come to read the electricity meters.

[2] It’s quicker than going to the entryphone.  Especially when it’s kiddies who ring the doorbell then run off.

apr_dbd_freetds driver

Over the past couple of days, I’ve been test-driving the APR DBD driver for FreeTDS, and getting to grips with it. Tonight I’ve reached the point where I’m (provisionally) satisfied that it’s working sufficiently for my client’s immediate needs, as well as for general webserver tasks such as SQL authentication and rewritemaps.

So, once I get round to uploading it, Sybase and Microsoft SQL join MySQL, PostgreSQL, SQLite and Oracle as databases supported or partially supported by APR/Apache.

The documentation sucks compared to either MySQL or PostgreSQL, but it’s a million times better to work with than the Oracle crap. And although these databases are not open-source, the FreeTDS source is eminently readable. That helps a lot!

There’s still lots to do before it can be described as a full-featured driver, let alone an adequately-tested one. But we have worthwhile progress. And a new feature that could be worth generalising to apr_dbd so all drivers can use it.

Regexp library chaos

Apache makes some use of regular expressions, for which it uses the Perl-compatible regular expression (pcre) library. And it bundles the library rather than making it a prerequisite: a decision which causes a range of problems (for example here) and sort-of fixed in an uneasy compromise.

APR doesn’t use any regexp support. So when I found myself using regexps in developing a FreeTDS driver for apr_dbd, I faced the decision of whether to use pcre or POSIX regexps. Two ideas led me to pcre:

  • PCRE is nice, and much more versatile/powerful than the (much older) POSIX regexps.
  • The application is within Apache, which already uses PCRE.

However, I then have to link PCRE into apr-util, which then links into an httpd instance linking its own pcre. This caused a whole bunch of trouble, and since I’m under some pressure of time, I abandoned the attempt to make it work and reverted to POSIX regexps. These in turn caused grief until I remembered that little thing, the REG_EXTENDED flag that POSIX requires to enable regexp memory. And a reminder of platform issues, when I debugged in gdb and saw just how different the regexp implementation on the Client’s Solaris platform is from my Linux development platform!

Not a big deal. Just a b***** annoying waste of time on what should be a trivial issue. Grrr …

The trouble with …

… turning off one’s mobile phone in the evening, is when one forgets to turn it back on in the morning, and misses what might be important calls.

Grrr …

The statistical likelihood of that is far lower than that of missing a spam call.  But then I won’t necessarily know the missed call was unwanted without spending time and effort on it.

Grrr …

This morning I don’t appear to have missed any.  So I can waste time blogging about it instead of faffing about with messages.

A new rain of spam

Yesterdays and todays news is that the ‘merkins have arrested one of their top spammers in Seattle. I don’t know how much difference this’ll make, but my understanding is that it’s one or two altogether different US states that give spammers a safe haven and could really make a big difference. Along with the world at large.

Here on the blog I’ve had a recent deluge of trackback spam pointing to something called “correctserver.com”. It’s a subtle one: I first saw it when I referenced an earlier post, and saw not just the one (legitimate) trackback, but a second one appearing simultaneously. I first took that for an innocent wordpress malfunction, then realised that the trackback from “[my post ]| Server software” was spam pointing to someone’s copy of my post. Since then I’ve had a number of them from the same spammer, and they get right through Akismet.

Today I just realised it’s more subtle than that. A week and a half ago, Danny Angus referenced my blog in an entry on his own. The first I saw of that was the trackback; then I saw it on Planet Apache. OK, fine, a legitimate trackback, right? Nope, it was only just this morning it showed up in my feed as [Danny’s entry]|Server software that I realised it didn’t link to Danny’s post, but to the spammer’s copy of it at correctserver.com.

A subtle and devious technique. WordPress admin and Akismet: I hope you’re listening!