IRC channel logs


back to list of logs

<nckx>ACTION just noticed that GWL is today's honoured ‘random package’.
<rekado>I’ll take this as my annual reminder to continue hacking on it before the HPC report comes along
<civodul>it’s an honor to be the random package of the day
<rekado>yesterday was the workshop and nothing went according to plan
<rekado>I had prepared a VM with the compiled Guix source tree, git, firefox, etc.
<rekado>but then people showed up with Macs
<rekado>I was told participants would have watched the 1 hour intro from last week but only one did.
<rekado>kinda derailed the whole thing
<rekado>and it was hard to keep on track because of audience questions
<rekado>was more like a really long Q&A with bits of presentations in between
<rekado>I’d be happy to hear how you organize hands-on workshops and how you keep them from falling apart
<rekado>for me the biggest difficulty is to keep it interesting for all participants in spite of their wildly differing backgrounds
<rekado>the organizers thanked me and were happy with the result, but I felt it was rather messy and not at all what I had in mind.
<rekado>some of the problems could probably be avoided by clearly communicating requirements, both for preparatory materials and for resources during the workshop
<rekado>but a common issue is that some people ask questions that totally derail the flow
<civodul>for hands-on session, we usually have VMs wit Guix preinstalled that people can use
<civodul>we ask for SSH keys beforehand and all, and invite them to optionally install it on their laptop, but it’s really optional
<civodul>(and there are Macs occasionally…)
<civodul>my experience is that there are often lots of questions, which is nice in a way
<civodul>but it means you have to be prepared to be derailed :-)
<civodul>in particular, i now spend more time on things like environment variables
<civodul>early on
<civodul>since that’s often a major source of questions
<rekado>i see
<rekado>we had one participant who actually had practical experience with Nix
<rekado>and some others to whom the whole concept of building software was rather foreign
<rekado>(windows admin I presume)
<rekado>the former works in a team that explores building the OS for their network switches with Nix
<rekado>sounds really exciting to me, to go beyond just the use of applications on legacy distros
<rekado>it always comes as a surprise to me how little admins know about environment variables
<rekado>there’s often a lot of confusion about how ‘commands’ become available to the shell
<rekado>but people lack an understanding of what components they are actually dealing with
<rekado>unaware of the shell, assuming that ‘Linux’ does something, etc
<rekado>for those participants it’s all just way too much new stuff to learn
<rekado>so by the time we get to the fun stuff they’ve already checked out
<rekado>(of course there’s also a good chance to run into long-winded but terribly vague ‘couldn’t we just…’ type of questions which essentially serve as a break for everyone else.)
<rekado>I got some questions about Spack and I think it’s much easier to give a fair, good answer about the pros and cons of Spack than it is to answer questions about Conda.
<jonsger>rekado: how many participants did that course have?
<civodul>was it HPC-oriented? mostly sysadmins?
<rekado>jonsger: initially there were … 2
<rekado>then I went to fetch a coffee with one of them, and upon my return there were about 15 or so
<rekado>civodul: it was a general ‘IT’ audience, not purely HPC
<civodul>and they’d heard about Spack?
<rekado>one of them, yes
<rekado>the IT4Science days had many talks from a management perspective (which is a foreign world to me for the most part)
<rekado>but in between there were also a few talks about scientific computing and working in the HPC mines
<rekado>I think the event scope was just a little too broad, so there’s no way of accommodating a mixed audience like this
<rekado>in early October I’ll have another presentation, this time in front of an audience of scientists who want to learn about scientific computing.
<rekado>that’s usually an easier audience for me to target
<civodul>you’re on every front these days!
<rekado>I feel like I’m making up for lost time :)
<civodul>heh, that’s good