12:04:45 <danshearer_> #startmeeting 12:04:45 <lumosql-meetbot> danshearer_: Meeting started at 2022-02-18T12:04+0000 12:04:46 <lumosql-meetbot> danshearer_: Current chairs: danshearer_ 12:04:47 <lumosql-meetbot> danshearer_: Useful commands: #action #info #idea #link #topic #motion #vote #close #endmeeting 12:04:48 <lumosql-meetbot> danshearer_: See also: https://hcoop-meetbot.readthedocs.io/en/stable/ 12:04:58 <danshearer_> #meetingname General Full-team Review 12:04:58 <lumosql-meetbot> danshearer_: Meeting name set to: General Full-team Review 12:05:05 <danshearer_> #here Dan Shearer 12:05:10 <rubdos[m]> #here Ruben 12:05:16 <Labhraich> #here Claudio 12:05:24 <BKJ621> #here Björn 12:05:32 <gabby_bch> #here Gabby 12:05:32 <danshearer_> #topic Reviewing from last meeting 12:06:02 <danshearer_> #item Gabby has got lots of benchmarks, so that's done, and we can more-or-less keep them coming 12:06:02 <lumosql-meetbot> danshearer_: Unknown command: #item 12:06:29 <danshearer_> #item Ruben did a great job, there is a working cluster, we have a little refining perhaps to do but basically it is there to make it sweat for us. 12:06:29 <lumosql-meetbot> danshearer_: Unknown command: #item 12:07:13 <danshearer_> oops 12:07:15 <danshearer_> I meant 12:07:40 <danshearer_> #info Gabby has got lots of benchmarks, so that's done, and we can more-or-less keep them coming 12:07:49 <danshearer_> #info Ruben did a great job, there is a working cluster, we have a little refining perhaps to do but basically it is there to make it sweat for us. 12:08:20 <danshearer_> #info Ruben carries over the action of committing BibLaTeX 12:08:26 <rubdos[m]> I sadly didn't do a lot of reading this week, as I planned, due to overhead, but that should be settled after today. 12:08:39 <rubdos[m]> Reading and BibLaTeX will go hand-in-hand, basically 12:08:54 <danshearer_> #info Björn sent out a reminder and here we are, so it worked. 12:09:29 <danshearer_> There's another section on benchmarking and stuff, but specifically as to actions from last meeting, I think we covered it. 12:09:53 <danshearer_> Anyone disagree or do we move on? 12:10:05 <Labhraich> remains silent 12:10:22 <danshearer_> Labhraich said something, so he wasn't silent 12:10:31 <danshearer_> #topic Benchmarking 12:10:35 <danshearer_> Oh boy 12:11:18 <danshearer_> #info Claudio anchored a really busy round of testing, bugfixing and improvements 12:11:42 <Labhraich> Yes, and as a result I didn't do much in the way of metadata... 12:11:54 <danshearer_> #info Benchmarking is now partly-concurrent (ie can be used on a shared cluster) as well as much more user-friendly 12:12:28 <danshearer_> #info There are now three people in the world who have run a benchmark and shared the results (Claudio, me and Ruben). 12:12:28 <Labhraich> It also now works on some non-Linux systems (NetBSD and FreeBSD), which was previously untested 12:13:12 <danshearer_> #info Benchmarking is now portable to more operating systems (NetBSD and FreeBSD), and does a lot more autodetection 12:14:28 <danshearer_> #info Dan has drafted a summary of our first results, intended for sharing with the small group of people with commit access to SQLite, LMDB or Fossil. 12:14:37 <danshearer_> Sooo 12:14:40 <rubdos[m]> Maybe worth mentioning that CPU information is also included in the benchmark file now 12:14:44 <rubdos[m]> pretty much automatically 12:15:37 <danshearer_> #info Benchmarking now automatically includes CPU info, and has a good try at including disk info. This is a work in progress, but extremely helpful in interpreting results. 12:16:01 <gabby_bch> I am working on the plotting app for the bencmarking results, I could put my R code on github layer today 12:16:04 <Labhraich> and we have a start of a tool and a procedure to combine results from people 12:16:17 <danshearer_> #info One thing we are aware of but have not started to solve is that it would be very useful to have a reasonably reliable Machine ID. There are many problems and opportunities with this idea. 12:16:42 <rubdos[m]> That'll be pretty difficult with the Kube system, although it probably does expose some magic env var for that 12:17:05 <rubdos[m]> standard hostname will be very random... 12:17:14 <Labhraich> If it has a CPU and/or a local disk... they both have serial numbers 12:17:23 <danshearer_> #info Claudio has created the start of a tool and a procedure to combine results from people. It works for trusted people driving it manually. 12:17:48 <danshearer_> gabby_bch, are you ok to take an action on putting up your R code? 12:18:20 <danshearer_> #action Gabby to commit R code and some sample results 12:18:30 <danshearer_> well too late, this moves fast :-) 12:18:31 <gabby_bch> ok 12:19:25 <danshearer_> rubdos[m], I see your point about machine Id being harder inside a shared cluster, however, that is also a fairly homogenous system from the point of view of high-level benchmarking like we are doing now. 12:19:45 <rubdos[m]> Yes, it'll be very possible, but also very kube-specific 12:19:53 <danshearer_> At the moment we are noticing big differences. I suspect usually per-node differences will be relatively small. 12:20:12 <danshearer_> So Kube machine id probably isn't important. 12:21:16 <rubdos[m]> exactly 12:21:23 <danshearer_> We did notice that there is a concept from DBUS and systemd and a package called nosh that creates a /etc/machine-id or /local/etc/machine-id in a standard form. But these are dependencies that won't be in your average container. So, that's for another day. 12:21:36 <Labhraich> We could just have a way to specify one manually for clusters or places where there isn't a reliable one. It's mostly so we can filter things "by system", and if we can get one automatically, great, if not... 12:22:17 <Labhraich> I'm not sure nosh helps at all. All it does is to copy information from elsewhere into /etc 12:22:35 <danshearer_> Ok, yes, I bet we can come up with a name for the entire cluster. If it says "VUB Cluster" then there will be someone responsible we can say "hey, look at this weird thing" and they will get all offended until it is fixed. 12:22:42 <danshearer_> Ok so in summary 12:22:58 <danshearer_> 1. Benchmarking is now super-excellently further forward than it was a week ago 12:23:34 <danshearer_> 2. We are within hours of presenting this to a group of 30-40 developers, and we are confident they can understand it and use it if they are motivated to do so. 12:23:59 <rubdos[m]> Like with the CPU_var_thingy we can just override those metadata pretty easily 12:24:09 <danshearer_> 3. There are still lots of results hiding inside this data, and more data needed, and quite possible improvements from the kinds of people we will be inviting to review. 12:24:25 <danshearer_> Anything more? 12:24:33 <danshearer_> I do want to add this to the notes... 12:25:43 <danshearer_> #info Dan wanted to make the point that the benchmarking that has been built is already at the top of its class. You cannot get anything comparable for other storage software, certainly not in a collaborative tool like this. Very well done. 12:26:32 <Labhraich> There's probably some more things benchmkark-filter can do to help making sense of results. But once we have a machine ID of some form that'll make things easier. A filter which returns data from an ARM32 with a SD card as well as from an Intel Xeon with a NVME will give out numbers which won't make sense 12:26:42 <danshearer_> #action Dan to send out benchmarking document v0.5 (or v0.6 if there is feedback from the team today). 12:27:08 <Labhraich> I haven't had a chance to read v0.5 today but will do that as soon as the meeting ends 12:27:41 <danshearer_> Ok, anything more on benchmarking? 12:28:12 <danshearer_> moving on 12:28:14 <rubdos[m]> Apart from Dan shooting more benches at the cluster as an #action 12:28:18 <rubdos[m]> I don't think so :) 12:29:00 <danshearer_> #action Dan to submit jobs to the VUB cluster and amalgamate the results 12:29:41 <danshearer_> Ok then 12:30:17 <danshearer_> we have our previous-agreed tasks and I'll summarise them in a simple note in a sec 12:30:32 <danshearer_> #topic Physical meeting 12:30:40 <danshearer_> this will be very brief. 12:30:45 <danshearer_> VUB says "yes" 12:31:02 <rubdos[m]> our groups says yes, and we'll force VUB to allow it ;-) 12:31:13 <rubdos[m]> potato potato 12:31:15 <danshearer_> I am working on a suggestion to arrive with Labhraich but I don't have a final answer on that 12:31:34 <danshearer_> rubdos[m] is eating root vegetables it seems 12:31:53 <rubdos[m]> carrots, indeed 12:32:00 <rubdos[m]> and potatoes too now I notice 12:32:09 <danshearer_> BKJ621 says yes and is cool with driving if not trains 12:32:13 <BKJ621> only question from me: when? + some info on practicalities 12:32:34 <danshearer_> gabby_bch is cool with jumping on a plane last we mentioned it, no sweat 12:32:40 <rubdos[m]> I'll gladly come and fetch people around if necessary 12:32:57 <rubdos[m]> that includes Charleroi airport 12:32:59 <rubdos[m]> within .be borders 12:33:06 <danshearer_> #info Ruben has offered car transport within Belgium eg Charleroi 12:33:07 <Labhraich> Trains from here seem impossible at the moment. And there's just been a new problem, a strike which means only 25% of the trains from here are running 12:33:23 <danshearer_> I am considering flying to Limoges and driving with Claudio 12:33:32 <Labhraich> I could be persuaded to that :-) 12:34:02 <danshearer_> I got stuck on accommodation. 12:34:10 <Labhraich> rubdos[m]: can you let us know what we need to get into Belgium? I assume vaccination certificate at a minimum 12:34:30 <rubdos[m]> They still change it every now and then, but I'd gladly find it out 12:35:31 <danshearer_> I looked it up and since I am the only one who is not an EU citizen and not coming from the EU it's a little stricter for me. Altho Scotland is not quite so aggressive as England (eg still have masks on public transport and in shops and nobody minds.) 12:35:44 <danshearer_> But even then nothing bad. I have had 3 shots. 12:36:03 <danshearer_> ooh I am going to try a new meetbot command 12:36:27 <danshearer_> #help Dan needs a hand to locate accommodation in Brussels 12:36:52 <Labhraich> danshearer_: I think the LMDB timing on test 8 is even worse than what I said earlier... multiplying the number of SELECTs by 5 multiplies the time by 3657 12:37:14 <Labhraich> Before you send that draft out... 12:37:24 <danshearer_> Labhraich, I have a size 5 or two to look at I think 12:37:39 <rubdos[m]> danshearer_: we can try the U residence, but that might be both expensive *and* unavailable 12:37:44 <danshearer_> I will add the word "extreme" in the summary points 12:38:01 <danshearer_> I looked up the U residence, and there are waitlists on all categories. It's term time. 12:38:07 <danshearer_> I tried to find youth hostels 12:38:15 <danshearer_> Because BKJ621 is so very young 12:38:25 <BKJ621> haha 12:38:33 <BKJ621> in mind 12:38:43 <rubdos[m]> Maybe look at fosdem-related pages for accomodation? 12:39:21 <danshearer_> If it wasn't a pandemic I would have already had offers of accomodation from locals I should think. But we can't have that fun at the moment. 12:40:37 <danshearer_> rubdos[m], for something like Fosdem I always hate having to travel from accommodation, so I get something close. But to you as a local, is there somewhere that is very practical and just 15 minutes away (especially when there are not ten thousand people looking for the same thing?) 12:41:10 <rubdos[m]> the problem is, VUB is enclosed in normal urban area 12:41:16 <rubdos[m]> so normal Brusseleirs live here 12:41:30 <gabby_bch> i'd say airBnB 12:41:49 <danshearer_> gabby_bch, I looked but I am not clever, so I probably didn't do a good jbo 12:42:11 <danshearer_> Let's get some actions on this 12:42:41 <rubdos[m]> #action Ruben to ask internally at VUB for places to stay 12:42:46 <danshearer_> rubdos[m], can you advise on transport? I know there's a bus, but I have only been on it when it was entirely unsuitable for a pandemic 12:43:11 <rubdos[m]> From where to where, exactly? 12:43:24 <rubdos[m]> Brussels public transport is pretty alright 12:43:25 <BKJ621> I have never done airbnb otherwise I have vounteered 12:43:50 <danshearer_> to == VUB. from == $anywhere nearby. 12:44:01 <rubdos[m]> 71 and 95 then 12:44:17 <danshearer_> ack on Rubdos claiming Brusselers don't want to sit on other passengers or squeeze them tight on bussses any more. 12:44:20 <rubdos[m]> there's also the semi-close metro 5 and the close tram lines 7, 25 and semi-close 81 12:44:43 <danshearer_> ooh I know nothing about metro near VUB. That sounds like a good way to teleport from say airbnb 12:44:46 <rubdos[m]> Well, I don't mind relatively packed busses too much in normal circumstances 12:44:48 <Labhraich> I've used airbnb once (unintentionally - it was booked via booking.com not via airbnb) and it was all right. Dan visited me there to have coffee :-) 12:45:04 <danshearer_> I've used airbnb lots of times. 12:45:10 <danshearer_> I can see why communities hate it. 12:45:33 <rubdos[m]> Try not to get an AirBNB in the west side of Brussels and transport should be fine 12:45:52 <danshearer_> Yes, in fact that is where I said to Claudio "I've got this thing I'm doing with SQLite" and he said "poor boy, you are doing it all wrong". 12:46:22 <danshearer_> #info Accommodation not on the West of Brussels and transport is good 12:46:38 <danshearer_> can someone volunteer to spend 20 minutes on airbnb.com ? 12:46:51 <BKJ621> yes 12:46:55 <Labhraich> I can see a number of hotels near VUB. THere are 2 types of them: "cost 1000 Euro a night" and "unavailable during that week" 12:46:55 <rubdos[m]> volunteer is invited to cross check addresses with me 12:47:31 <rubdos[m]> That sounds like Etterbeek and Oudergem, you should probably try hotels in Elsene instead, those will vary more in price 12:47:31 <danshearer_> #action Björn too look up AirBNB in Brussels, checking with Ruben, and looking for any week in March if dates are difficult 12:47:46 <Labhraich> But we really need to agree on a day because the rest of the planning really depends on that 12:47:50 <danshearer_> rubdos[m], Elsene is no good that week 12:48:04 <danshearer_> I even tried the Dutch name but it was no better. 12:48:09 <Labhraich> Yes, I did look at Elsene. That's the one "unavailable that week" 12:48:25 <rubdos[m]> Ah damnit. 12:48:32 <rubdos[m]> Ixelles is the French name 12:48:41 <rubdos[m]> Excuse my French. 12:48:43 <danshearer_> How about "the week in March where accommodation is easiest in Brussels?" 12:49:01 <danshearer_> Oh I see 12:49:40 <rubdos[m]> I'm one of those annoying people that use the unused Dutch names. 12:49:46 <danshearer_> #info parking a car is difficult/expensive in Brussels, but allegedly not a problem further out 12:49:57 <Labhraich> Actually I had a hotel in Ixelles/Elsene during FOSDEM, so I've looked there first 12:50:00 <rubdos[m]> I know the free spots 12:50:04 <rubdos[m]> for parking 12:50:05 <danshearer_> rubdos[m], booking.com uses Elsene. Which sounds French to me. 12:50:08 <rubdos[m]> but don't come with large vehicles. 12:50:16 <BKJ621> How many days in Brussels? 12:50:28 <BKJ621> What is a large vehicle? 12:50:29 <rubdos[m]> danshearer_: Elsene misses at least two accents in order to be a French name ;) 12:50:30 <danshearer_> #info Ruben knows the free spots for parking a small vehicle 12:50:45 <rubdos[m]> SUV is doable, anything over that is at least annoying to park 12:50:51 <Labhraich> Êlsèné? 12:51:03 <rubdos[m]> Don't do motorhomes. :-) 12:51:12 <danshearer_> No, rúbdòs 12:51:17 <rubdos[m]> Êlsèné or Êlsénè inded 12:51:30 <rubdos[m]> I see you all have compose keys preinstalled. Good. 12:51:44 <danshearer_> Björn said I had to 12:51:55 <Labhraich> Since I use a UK keyboard, it would be impossible to write any French without compose keys 12:51:58 <danshearer_> So I said yës 12:52:33 <rubdos[m]> åwsöme 12:52:45 <danshearer_> ok maybe I am familiar with compose keys 12:53:25 <danshearer_> do we agree that accommodation drives dates in march? 12:53:52 <BKJ621> We must agree on #of days too 12:53:58 <danshearer_> #accepted We meet in march when accommodation is possible and Björn/etc will tell us that 12:54:30 <Labhraich> Yes, number of days is as important as start date... 12:54:44 <danshearer_> #motion I propose we say the minimum number of days to meet in Brussels is two, with no maximum 12:54:44 <lumosql-meetbot> danshearer_: Voting is open 12:54:56 <Labhraich> It does feel like too long a journey for 1 day of meeting (unless somehow the pandemic disappears completely and we can get there in full comfort again) 12:55:19 <Labhraich> #vote +1 12:55:23 <danshearer_> #vote +` 12:55:28 <danshearer_> #vote +1 12:55:32 <rubdos[m]> #vote +1 12:55:49 <BKJ621> #vote +1 12:55:51 <gabby_bch> #vote +1 12:55:56 <danshearer_> #close 12:55:56 <lumosql-meetbot> danshearer_: Motion accepted: 6 in favor to 0 opposed 12:55:57 <lumosql-meetbot> danshearer_: In favor: Labhraich, danshearer_, danshearer_, rubdos[m], BKJ621, gabby_bch 12:55:58 <lumosql-meetbot> danshearer_: Opposed: 12:56:18 <Labhraich> Interesting. The bot lets people vote multiple times 12:56:36 <danshearer_> I though I made an invalid vote 12:56:46 <danshearer_> finished on meeting in Brussels 12:56:52 <BKJ621> should we plan for at least 2 full days in Brussels? 12:57:00 <danshearer_> I think final topic is reminder on agreed jobs 12:57:00 <Labhraich> Have we discussed when? 12:57:25 <BKJ621> Dan indicates some "marketting" activity too 12:57:33 <danshearer_> Labhraich, I thought we said accommodation had to drive it? We picked a week but there is no room in Brussels that week. 12:58:24 <Labhraich> OK 12:58:32 <danshearer_> This is not my favourite activity, by which I mean, I am not very good at it. 12:58:36 <BKJ621> sorry: I missed the motin. I am happy with that 12:58:46 <BKJ621> motion 12:58:55 <BKJ621> Keyboard is bas 12:59:04 <danshearer_> I just said "#accepted" which I can type at any time. Even if in fact nobody really accepted it :-) 12:59:34 <danshearer_> #accepted We meet for a minimum of two days in Brussels, with no upper limit. 12:59:52 <danshearer_> BKJ621, you have a very good point 12:59:57 <danshearer_> I really did say that. 13:00:13 <danshearer_> I have warned NLnet that there is a meeting and they are invited 13:00:31 <danshearer_> I want to do a presentation open to the public, including security people who work in Brussels 13:00:54 <danshearer_> None of which I can really do without a date 13:01:12 <danshearer_> #action Dan to coordinate marketing activities with NLnet, VUB and others once a date is agreed 13:02:00 <danshearer_> that meeting would be say 1 hour at the end of the two days, or something. Perhaps 2 hours, 1 being super technical and good for VUB staff, and 1 being "what this stuff means for society" 13:02:23 <danshearer_> I note that time is moving on. 13:02:42 <danshearer_> are we finished with #brussmeet? 13:03:01 <BKJ621> yes 13:03:07 <Labhraich> Nothing else from me 13:03:40 <rubdos[m]> me neither 13:03:45 <danshearer_> #info Dan proposes 1 hour public presentation on LumoSQL super technical and good for VUB staff, and 1 hour for security people and so on "what this stuff means for society" 13:03:59 <danshearer_> #topic Reminder to ourselves about jobs in progress 13:04:27 <danshearer_> #info Ruben is doing academic reading and noting papers, with BibLaTex commits to LumoSQL tree 13:04:44 <danshearer_> #info Gabby continues on graphs and also a docs task active 13:05:09 <danshearer_> #info Claudio is back on metadata, which is a requirement for hidden columns, encryption etc 13:05:55 <danshearer_> #info Dan is on Lumions/reading too; coordinating feedback with other project authors, looking at marketing for March Meeting and other things. 13:06:12 <danshearer_> #info Björn needs to talk to Dan about NLnet claims 13:06:26 <danshearer_> Anything else? 13:06:39 <BKJ621> nothing from me 13:07:04 <rubdos[m]> Sounds fine to me! I'd love to hear the distinction between "security people" and "super technical staff", but that can wait until another time 13:07:54 <BKJ621> And where does the NLnet people fit & other people w/ money 13:08:14 <danshearer_> yes ok, I will followup on those items 13:08:26 <danshearer_> #action Dan to circulate notes on marketing audiences 13:08:41 <Labhraich> I have a few things to add about the draft document, which I'll do in the next half an hour or so. After that I really need to concentrate on metadata and treat benchmarking as a distraction 13:08:56 <danshearer_> #action Dan and Björn to discuss funding 13:09:03 <danshearer_> ok! 13:09:05 <danshearer_> we are done! 13:09:09 <danshearer_> Thankyou everyone 13:09:22 <danshearer_> #closemeeting 13:09:22 <lumosql-meetbot> danshearer_: Unknown command: #closemeeting 13:09:27 <danshearer_> #endmeeting