Keysigning

We had a very small keysigning last Wednesday at ApacheCon (thanks Jean-Frederic Clere for organising it). I exchanged identity details with about 20 others, many of whom I already know.

Today I got around to digging up my list of details, and signing 11 keys I hadn’t already signed from some earlier event. If you were there and I got your details, my signature on your key should now appear on the keyservers.

ApacheCon last day

OK, most of it hasn’t happened yet.  I’m sitting in the ASF pioneers panel, a non-techie event to celebrate 10 years of the ASF, and lighten the day.  Just checked out of the hotel and left the luggage with concierge.

This afternoon sees the best Apache HTTPD sessions.  I’m not giving any of them, but I’m down for the easy task: chairing two sessions.  May miss out on blogging them, as we’ll be off to catch the ferry later this afternoon.

After two days of pissing rain, the weather has cleared up and the sun is shining.  Poor ol’ Sarah got soaked twice while touristing the city during the day, but it looks like she’ll at least be spared travelling with wet stuff.

Language abuse bites back

In the early days of Tomcat, they settled on an architecture wherein tomcat is fronted by an HTTPD proxy, which would also serve the application’s static data.  That makes sense, not least from a performance point of view.

But it gives rise to a by-product.  Tomcat folks come to use the word static to refer to anything served by httpd and not tomcat, regardless of whether it is in fact static or dynamic.  A minor irritation, but not one on which I’ve hitherto felt the need to rant.

Today that little abomination bit back, when Jean-Frederic Clere used it in his talk on proxying/clustering/loadbalancing solutions for tomcat.  An audience member took the word at face value and questioned him on a usage that just didn’t make sense.  Oops!

(no disrespect to Jean-Frederic – a much-respected developer, and since English isn’t his native language, he has no strong reason to feel the abuse).

ApacheCon day 1

Today was the first day of the main conference.  It mostly feels very alive this year, though that may or may not have any basis in reality.  Today was also hugely improved by the fact that we got a very good lunch, and my stomach was sufficiently recovered to enjoy it.

There were no talks today directly relevant to my work, so I thought I’d try some of the hadoop track, to learn a little about one of the ASF’s more exciting projects.  Unfortunately Owen O’Malley’s introduction in the first slot was too packed to get in, so I went elsewhere.   I did get to see Olga Natkovich’s[1] talk on pig after lunch, but that was all for hadoop.

The other interesting thing was in the business/social track.  I made a point of going to see Gianugo Rabellino’s talk: he’s a fine thinker and speaker, and always both interesting and entertaining.  I hadn’t planned to, but I stayed on to hear Paul Freemantle on WS02′s business model: fascinating because it had so many echos of my own experience in, and WS02 appear to be making a success of it.

In the early evening was keysigning.  Not so many this year: it wasn’t really adequately advertised.  But I’m certainly exchanging signatures with a few new folks, which is always useful.

[1] Natkovich?  Have patronymics gone unisex/politically correct?

Day Zero

The second day of ApacheCon week was the hackathon.  Having had breakfast at subway sandwiches next door to the hotel on Monday, we tried again, but they had no bread.  I managed to get a dicky tummy after no breakfast, and was running in and out a bit during the day, but that pretty-much cleared up by the end of the day.

In the hackathon, we had a chat around APR.  Is there really a good reason for APR-UTIL to be a separate library (nope)?  The work we’ve done recently to separate out big dependencies like the DBD libs is right – we don’t want to load a library we’re not using.  One option is to follow this direction further towards full modularisation.  And there seems little benefit in wrapping LDAP at all: so far as anyone knows, HTTPD is the only user of apr_ldap, and HTTPD can just clone what it needs from that code.  Some of that discussion has already moved to the mailinglist, to get the views of a wider community.

Tuesday being also Sarah’s birthday, it was a lovely coincidence that de Nederlandse Opera were performing Cosi fan tutte in he evening.  The production was a little strange: set in modern times, opening on a beach.  Despina was – at times – something straight out of the red light district – and had the figure for it!  By the end of the first act we were well-confused, but by the end of the evening it worked very well for us, and we much enjoyed it.

Coming out of the opera, we managed to get somewhat lost, before making our way to the Damrak and a reasonably decent late-opening pizza place that was sit-down rather than a dodgy takeaway.  Didn’t join apachecon folks in the hotel bar afterwards, as I was still not exposing the tummy to booze.  But I think I can draw a line under that today.

I think it’s now time to head down to the action!

BarCamp

Day 1 of ApacheCon Week was BarCamp.  And it was lots of fun!  In fact I preferred the friendly, informal style to that of the main conference: it’s more engaging, and it caught my interest in subjects that would otherwise probably have passed me by (such as Mahout, or a Lucene search), as well as more obvious candidates for my attention like open geodata, and apachecon classic themes like awareness of FOSS in education.

My “apache-helpdesk” session on support ecosystems for open-source projects was scheduled last, so we could have DrBacchus in (he was busy with training for most of the day).  He and pctony are the people present at ApacheCon who I already knew to be leading lights of that ecosystem, and it would’ve been a shame to run the session without both of them.  The session generated some discussion, but no great input from other projects, and I suspect the reality may be that httpd indeed leads the field.  A few slides I used to prime the discussion are here.

Only fly in the ointment was having to faff off into town to find our own lunch.  But that was amply compensated when I got the text from Sarah saying she was indeed – as hoped – coming to join me.  After meeting her at the station and dumping her luggage back at the hotel, we went for a wander around town, ending up in a little thai restaurant where we enjoyed an utterly gorgeous meal.  Yum!

Grumble

So once again I’m in Amsterdam.  My hotel room is hot and very noisy, but at least it’s a comfy bed and there’s internet access.

Yesterday started off just fine.  Flight was half-empty, but on a propellor-driven plane which was noisier inside than I’ve encountered before, and vibrated quite a lot.  I had an excellent view of the left propellor, and it created some strange visual effects: a very static shadow whilst on the tarmac, and a rainbow shadow on cloud.  Bit of a drama getting to the plane, too: they put us on a bus but it wouldn’t start, so they brought another bus.  The distance the bus had to take us was 1 minute’s walk!

There were quite a few Apachecon folks around when I arrived in Amsterdam, so after a shower I joined them in the bar.  Later we went out for what was originally going to be an indian meal.  But then they ended up going into a b***** steak house.  Faced with a choice of going in or going off alone, I did the wrong thing and went in, to eat their only vegetarian option: a deeply uninspiring platter amongst which the only bit with any taste was the two one-inch-diameter mushrooms.

With a start like that, things can only get better.  Can’t they?

ApacheCon fun

Well, I’ve left booking apachecon far too late this year.  Results: mixed, somewhat interesting.

Travel is a pain: I was unable to book my usual dutchflyer journey (inclusive train+ferry ticket for a pleasant journey at a fantastic price).  So I’ve done a very bad thing, and booked a single flight out (13:30 from Exeter – and it seems I have to take a taxi to get there in time for it).  For the homeward journey I was fortunately still able to book a dutch dutchflyer-equivalent single journey by ‘phoning Stena Line’s dutch booking number.

Booking ApacheCon was slightly irritating, because it insisted that my ASF-members discount code wasn’t valid.  So I declined to pay online, and will argue that point with them before they get my creditcard.

Most interesting was the hotel booking.  I booked the Moevenpick – the conference hotel – as a line of least resistance, despite my very bad experience with them two years ago.  Following the apachecon link to booking at the apachecon price, it told me nothing was available.  But it did give me a price for general booking, that was €242 cheaper for five days!  It’s still 50% more than I paid for the vastly nicer Renaissance hotel in Brussels for FOSDEM last month, but a worthwhile saving on apachecon’s rates, or the last three years ApacheCon hotel prices (both Dublin and Amsterdam being hideously overpriced).

Expect some apache-focussed blog entries over the next week

ApacheCon Submissions

I consider it too far to go from here to New Orleans for the forthcoming US ApacheCon, unless my employers were not only to pay but also to insist.

But I’ve got around to submitting two proposals for Amsterdam next March:

• Virtualising the Webserver
• mod_rewrite is obsolete

Dunno if either or both will be accepted, but I think they should be worthwhile topics. Both involve presenting thoroughly up-to-date and topical work: the first addresses a longstanding unfulfilled need for hosting companies, while the second is about developments that simplify Apache configuration, and could look like moving from editing a sendmail.cf by hand to a modern easy MTA like postfix.

One I thought about but didn’t put in was the Sun web stack[1]. As a topic it feels rather too commercial to me, and I suspect also to other Apache folks. And I wouldn’t be the right person to talk on that: it wants someone with far better presentation skills. I also omitted the modules tutorial I’ve given in some previous years: that was always too much of an organisational nightmare, requiring a minimum number of signups, which was commonly reached only after the notional closing date for signing up …. you get the picture.

[1] I should stress that none of my colleagues suggested I should speak on the subject: it was purely my own bad idea for being a corporate type.  Fortunately, bad ideas are only a problem if you hold on to them to the point of acting on them.

ApacheCon CFP

The Call for Papers for ApacheCon EU 2009 has just opened.

I have yet to decide what to submit for it, but I anticipate at least (and probably exactly) one submission based on current work at Sun.  Watch this space.