Monthly Archives: September 2020

ApacheCon@home

This year, for obvious reasons (covid), ApacheCon is taking place entirely online. Today was the first day. So what was it like?

Well, obviously it’s not a Jolly in a nice hotel in an interesting location, as the best events have been. Does that detract from it? Well of course those are part of the magic of the best ApacheCons – above all Budapest, where both the hotel and the city were fantastic. On the other hand, the money saved could buy quite a decent week’s holiday somewhere of my choosing! Better to focus on what did or didn’t work well in terms of presentations, communication, networking.

The economics of online worked nicely for lots of people, bringing in several thousand attendees compared to a few hundred at a “normal” event. A poll suggests that eightysomething percent of those thousands are attending their first ApacheCon: evidently a lot of people find it a lower hurdle (as I do). So we’re embracing a much bigger community, which is fantastic – so long as we don’t disappoint.

The times also worked nicely for us in European (and indeed African) timezones. While Americans, Asians and Antipodeans had it taking much of their nights at one end or t’other, here it opened at 09:30, and ended with BoF sessions at 21:00, much like a regular ApacheCon. The benefits of being in the middle of the world’s inhabited timezones!

After one or two initial glitches – a very short learning curve – the technology worked well. Presentations were clear, with the presentation window split to show the presenter, his/her screen, and a text chat window in separate panes: pretty-much ideal. “Corridor” action was, I thought, less successful, but then I’ve long found text chat easier to work with than face-to-face, which is why I don’t even have a webcam and audio system on my desktop ‘puter. Text chat there was a-plenty on every possible topic, but then we don’t need an organised event to benefit from that.

In terms of contents, the programme was easily as good as any I can remember. I enjoyed and was inspired by a number of talks, including some not merely on subjects but on projects with which I had no previous familiarity. In fact I think it worked rather better than sitting in a conference room, and I found it easy to stay alert and focussed, even in that after-lunch siesta slot when it can be hard to stay awake.

A major theme this year is the rapidly-growing Chinese community at Apache. In recent years it’s moved on from a handful of individual developers contributing to Apache projects, to quite a number of major projects originating in China and with Chinese core teams coming to Apache. Sheng Wu – a name I’ve hitherto known as a leading light of the Incubator and also lead on one of those projects – gave a keynote on the subject.

I don’t recollect when we had the main discussion of the language issues of Chinese communities coming to Apache, but some of these are now fully bi-lingual, with English being ultimately the official language but Mandarin also widely used. Mandarin was also used in a few of the morning’s talks – morning being of course the eastern-timezone-friendly time of day (and there was also a Hindi track). One Chinese speaker whose English-language talk I started to listen to proved hard to follow, and I found sneaking out unobtrusively a minor benefit of the online format!

The success of Chinese projects coming to Apache was demonstrated by two of today’s most interesting talks – by Western speakers (one American, one German) who don’t speak Chinese, but have become members of the respective core developer communities by virtue of participating. One was about the project itself, but Julian Feinauer’s talk was specifically focussed on the community: how a bi-lingual community works in practice (a question on which I’ve mused before, for example with reference to translations of my book, and regarding nginx). Answer: it’s working very well, with both languages, with machine translation to help “get the gist”, and with bilingual members of the community. And there are gotchas, when an insufficiently-comprehensive translation leads to confusion.

Summarising the Chinese theme, I think perhaps Sheng Wu’s keynote marks the point I dreamed of when I wrote the preface to the Chinese translation of my book.

Congratulations to Rich Bowen and his team on adapting to the circumstances and bringing us a fantastic event! More to come, about which I may or may not blog.

Bloat

Details still to be finalised, but I have a sponsor for a new task that’ll help improve security on the Web for all of us.  All to be open-source.  I expect I’ll blog details about it in the not-too-distant.

This has helped me feel more motivated than I have done for quite a while: even before covid and lockdown I was far from my most productive.  This week I’ve been updating infrastructure for the work, with a new raspberry pi on order, and many hours of huge dist-upgrade on my desktop box (running Debian).

The update appears to have run smoothly, though at one point the box wouldn’t wake from suspend (or from screen-blank) correctly and needed a hard power-cycle followed by recovery.  Touch wood: apart from that transient glitch, nothing worse happened (yet) than my (utterly unimportant) desktop background going away (now replaced).

What struck me though was the sheer bloat of a modern system, that bit me before the update.  In preparation I made a much more complete backup than my usual (which is, only data that matters and where mine is an original – so excluding huge areas like working git and svn repos that get committed upstream).  This time I tried to backup the whole of /home, and came a cropper: the compressed tar archive exceeded the 4Gb maximum file size on the USB stick I was using!

OK, how to reduce it?  There’s a fair bit of low-hanging fruit: for example, build directories where “make clean” removes much, and the Downloads directory whose contents are always dispensible.  But the real eye-opener was the caches for various applications – and not just obvious suspects like web browsers!  Some of it years old: for instance, an RSS reader I haven’t used for at least two or three years.  And in mailers, a huge discrepancy between Evolution and Claws mail: the former being many times larger, and appears to keep not just a copy of my archives but even long-deleted spam.  I cleaned a lot of cache, and removed some caches (including Evolution – which is in any case far too sluggish) entirely.

We’re spoiled rotten by the size of even the smallest modern storage, and get sloppy in accumulating junk.  df now (after big cleanup and dist-upgrade) shows my disc 18% full, so I’m under absolutely no pressure, but I shall nevertheless take this as a wakeup call to configure caches to limit their size and lifetimes for ancient entries.

Is it only us greybeards who ever give a thought to our digital hygiene?