11:57:43 <danshearer1> #startmeeting 11:57:43 <lumosql-meetbot> danshearer1: Meeting started at 2022-03-04T11:57+0000 11:57:44 <lumosql-meetbot> danshearer1: Current chairs: danshearer1 11:57:45 <lumosql-meetbot> danshearer1: Useful commands: #action #info #idea #link #topic #motion #vote #close #endmeeting 11:57:46 <lumosql-meetbot> danshearer1: See also: https://hcoop-meetbot.readthedocs.io/en/stable/ 11:58:56 <danshearer1> #meetingname lumosql.20220304.1300 11:58:56 <lumosql-meetbot> danshearer1: Meeting name set to: lumosql.20220304.1300 11:59:03 <Labhraich> #here Claudio 11:59:12 <danshearer1> #here danshearer1 11:59:13 <BKJ621> #here Björn 11:59:37 <danshearer1> please - can you #here your nick. The bot isn't that clever for assigning tasks. Yet :-) 11:59:52 <danshearer1> as in "#here Labhraich" 12:00:06 <danshearer1> and welcome alles 12:00:06 <Labhraich> According to the meetbod docs, you say "#here (your real name)" and that associates the nick with the name 12:00:23 <gabby_bch> #here gabby_bch 12:00:24 <danshearer1> hum ok then I don't know what I'm talking about. 12:00:35 <Labhraich> The meetbod already knows the nick 12:00:39 <Labhraich> "meetbot" 12:00:46 <danshearer1> Ignore what I said. I have raised two bugs/suggested code changes in the bot including improving this bit. 12:00:47 <danshearer1> anyway 12:01:11 <danshearer1> #topic Introductory stuff 12:01:33 <danshearer1> #info Ruben is driving and therefore not here. But he posted notes earlier 12:02:53 <danshearer1> #info Ruben will commit his bibliographic data to doc/references later 12:03:04 <danshearer1> #info Ruben's summary is " ABE is not ready for the feature set we want to support. My next goal with this document is to find out which subsets of the feature set we can support" 12:03:20 <danshearer1> I don't want to speak any more on Ruben's behalf but I think that's fair enough. 12:03:44 <Labhraich> If one has to summarise the whole morning in 3 lines, it looks fair enough to me 12:04:32 <danshearer1> Today I think our topics are: Brussels in April; Next step in docs/benchmarking ; internals 12:04:45 <danshearer1> anything more in the introductory bit before we get to Brussels? 12:05:24 <danshearer1> ok it seems not 12:05:34 <danshearer1> #topic Brussels Meetup 12:06:02 <danshearer1> #info Björn has sent an email with details. We need to agree and then do the actions. 12:06:13 <BKJ621> exactly 12:06:15 <danshearer1> BKJ621, off you go, you have the floor of this circus tent 12:06:49 <danshearer1> No need to repeat the email. What are the decisions. 12:07:20 <BKJ621> Can we agree on the proposal? Any problems identified? 12:08:27 <gabby_bch> I've found a very cheap flight Liverpool to Brusselles 12:08:54 <danshearer1> I think the first question is: can everyone get to Brussels April 3-6 inclusive 12:09:26 <BKJ621> anyone that cannot? 12:09:39 <danshearer1> ruben said he is fine with that 12:10:01 <BKJ621> Do we have a venue at VUB secured? 12:10:05 <danshearer1> ok then we are agreed. 12:11:03 <Labhraich> It would probably be easier to arrive on Apr 2, but I suppose Apr 3 is doable 12:11:26 <BKJ621> just minor clarification, last activity on Apr6. Checkout on Apr7 for those at the hotel 12:11:29 <Labhraich> Will need to check the trains (... and the works on the line!) 12:12:06 <danshearer1> BKJ621, can Labhraich have another night if that makes sense practically? No point in enforcing misery in travel to save one night. 12:12:36 <BKJ621> Yes. I believe so 12:12:43 <Labhraich> It's just that... when I looked up the trains it had a message about works until Apr 15 - but doesn't appear to affect Saturday or Thursday trains 12:12:59 <danshearer1> Labhraich, ok but Björn just solved that problem. 12:13:02 <Labhraich> So I might find it better to avoid travelling on Sunday 12:13:25 <danshearer1> gabby_bch, are you good for arriving Sunday or so and leaving Thursday or so? 12:13:26 <BKJ621> I spent 4y on the train to Stockholm -- you never know 12:13:49 <danshearer1> BKJ621, that is the slowest train I have ever heard of. You must have made a lot of friends during those 4 years. 12:14:07 <Labhraich> From where to Stockholm? 12:14:08 <gabby_bch> I can arrive late evening on the 3rd and leave on the 7th 12:14:28 <BKJ621> 4 turns around the globe (Göteborg - Stockholm) 12:14:48 <Labhraich> Let me check the trains before booking the hotel - I haven't excluded arriving on the 3rd, just have to check the options 12:15:04 <danshearer1> Labhraich, nevertheless, you can bracket those dates one way or another? 12:15:39 <Labhraich> Yes, I have nothing planned. I might have to get to the station to buy the ticket, but with a month of time I'll find the ticket office open at some point... 12:15:59 <BKJ621> So Björn is ok, Clauido is ok, Dan is ok, Ruben is ok, Gabby is ok, Tom is ok?? 12:16:17 <Labhraich> (Last time I did that because their online system refused to work, the ticket office spent 20 minutes swearing at it and saying "ne marche pas" but never mind that) 12:16:44 <BKJ621> Just must keep hotel stay for Clauido a bit flexible 12:16:54 <danshearer1> BKJ621, Tom is a local and quiet here. He can raise an interrupt at any point. 12:16:59 <danshearer1> #accepted We meet in Brussels, arriving at latest on Sunday 3 April and leaving earliest Thursday 7 April 12:17:11 <Labhraich> I'll let you know - obviously if I manage to book a ticket for Apr 3 on one of the trains guaranteed to run, that'll be best 12:17:58 <danshearer1> gabby_bch, Björn or I can book the flights if that will help, to save booking then reimburse. As you choose. 12:19:04 <danshearer1> ok so we all understand that we need to tell Björn when we are arriving in Brussels so he can book the hotel rooms. 12:19:09 <BKJ621> Better if you book yourself. I will work NLnet to secure if we need to split payout into 2. tickets first the rest later 12:19:27 <gabby_bch> I'd rather book them myself to avpid sending tickents back and forth 12:19:33 <danshearer1> BKJ621, you are great 12:20:21 <danshearer1> #accepted We book however we want and then tell Björn (a) the dates so he can book rooms and (b) tickets so he can handle reimbursements 12:20:38 <BKJ621> I will reserve hotel rooms but you pay. Reimbursed afterwards. A must to provide receipts etc for major stuff 12:20:45 <BKJ621> Exactly!! 12:21:39 <BKJ621> This will be fun 12:22:12 <BKJ621> I think we are done with that part of the Brussel question 12:22:32 <danshearer1> Good. 12:23:14 <danshearer1> VUB have promised us a lab to be in. This will be our convenient HQ 12:23:51 <BKJ621> I guess we need some larger space for the presentations 12:24:05 <Labhraich> OK, there are a grand total of 3 seats left on guaranteed trains on Sunday 3/4... so as long as I book soon 12:24:16 <danshearer1> We have to trust Ruben and his boss An that it is suitable as our hanging out space for getting individual work done as well as working together. 12:25:09 <danshearer1> For presentations, VUB have plenty of time to book appropriate lecture theatres. And I heard informally that even if VUB do not, then their sister next door who host FOSDEM can :-) 12:26:07 <danshearer1> #action Dan to confirm to VUB about room requirements as soon as we know. Certainly one lab as promised and one good sized presentation space. 12:26:31 <BKJ621> Labhraich you tell me if you need a room from Apr2 already or not 12:26:41 <BKJ621> when you know 12:27:10 <Labhraich> I'll know tomorrow at the latest - if I need to go and buy the ticket at the station I'll go there tomorrow first thing 12:27:53 <danshearer1> which brings us to the next topic. I made some suggestions for talks and their audiences. NLnet and VUB between them want to bring various people along. And as to speakers, I am a boring talk-machine at any time and Ruben is keen as well. Volunteers for anything else welcome but not insisted on. 12:29:20 <danshearer1> To save any logging in etc, here are my suggestions in text: http://paste.c-net.org/LandoCheeks 12:29:26 <Labhraich> I'm not sure I have anything to talk about which would be of more general interest. I could go on for a bit about hiding data in a sqlite database but that's about all.. 12:30:18 <Labhraich> I suppose if I can get my head out of vdbe I might be able to think of something else 12:30:21 <danshearer1> Labhraich, so lets just say its a bit like an EdLug used to be... for those that would like it, some SQLite internals chat "I could go on for a bit about hiding data in a sqlite database but that's about all.." 12:31:15 <danshearer1> Maybe more of an informal collective chat about SQLite database internals, and if you think nobody cares about that I think you are wrong ;-) 12:31:15 <Labhraich> But I've only ever gave April 1st talks at the EdLUG - oh wait, this is going to be the 1st week of April :-) 12:31:58 <danshearer1> Exactly. And now you live in France, which has poisson d'avril, so in this case it would be like the April Whale, one giant week-long crazy talk. 12:32:15 <danshearer1> All right. 12:32:43 <Labhraich> OK then, I'll give an introductory talk about how I would go on about hiding data which vdbe doesn't notice (but also doesn't delete) and if people want to go from that to sqlite internals they'll have to say something about it 12:33:21 <Labhraich> Might even have a quick introduction why I am hiding the data there 12:34:09 <danshearer1> If I was in the audience and didn't know anything about the topic, I would definitely put my hand up and ask "So, are you saying you have implemented a way of subverting trillions of databases without anyone noticing, potentially inserting data in a way that is then revealed later by some totally innocent action such as ALTER TABLE?" 12:34:33 <danshearer1> So basically yes, you are a scary person 12:35:10 <danshearer1> Ok well I know Ruben can talk a lot about his favourite topics of crypto and he is starting to feel like he will have something to say one month from now. 12:35:19 <danshearer1> And I have some bigger picture thoughts. 12:35:35 <danshearer1> and the only person present here who has something to actually see at a glance is Gabby. 12:35:42 <Labhraich> Remembere my QR code I hid near your house years ago? The one which scanned to "Why are you reading this?" but if you looked at it the right way it gave GPS coordinates of something else? It's not like I've started hiding data recently :-) 12:36:05 <danshearer1> remembers well that Labhraich likes geocaching 12:36:47 <Labhraich> Right, then, I'll try to make up some form of title which doesn't give away "hiding data in trillions of databases" too soon 12:37:47 <danshearer1> So Gabby, would you like it if there was a section in one of my talks where you get to show some pretty graphs a bit like http://r.lumosql.org:3838/lumo-index-2.html ? 12:38:10 <danshearer1> You get to say if you like or hate the idea :-) 12:39:29 <danshearer1> So... 12:39:39 <danshearer1> I need to be able to advertise some talk slots. 12:39:51 <gabby_bch> um I don't think I sould present anythong, you can of course use the graphs in your talk 12:39:52 <danshearer1> How about the afternoon Wednesday 6th April? 12:40:03 <danshearer1> gabby_bch, np 12:40:33 <Labhraich> I suppose with the right encouragement (*) I could possibly figure out a way to talk about how we do benchmarks 12:40:44 <Labhraich> (*) I hear they have good beer in Belgium 12:40:52 <danshearer1> that's just waffle 12:41:24 <danshearer1> Ok good. 12:42:21 <danshearer1> #action danshearer1 to work on talk abstracts, and checking VUB can host 6th April, and then advertise 12:42:26 <Labhraich> What language are we going to give the talks in? My Dutch is rusty, but I can offer English, French, Italian. I can also offer Russian but right now I'd rather not 12:42:44 <danshearer1> ahha! I remember now, it isn't #here that needs the nick, it's #action. Ok, I was mis-remembering. 12:43:16 <Labhraich> Yes, but if there was a #here before, meetbot can use it to map real names to nicks and vice versa 12:43:43 <danshearer1> yes, and that's the meetbot bug. I just forgot where the bug was. But I have reported it and suggested a fix :-) 12:44:53 <danshearer1> Talks to be in English with French bits where they make sense. Anyone invited for beers or academic chats is probably going to find French helping a lot. 12:45:29 <danshearer1> (This is Brussels right, where the fastest-growing languages iirc are Spanish and German and we aren't doing either :-) 12:45:45 <danshearer1> ok are we finished with the Brussels topic? 12:45:57 <Labhraich> You never know. At FOSDEM at some point I have been volunteered to translate from Italian to French while conversing in German with somebody else. A bit confusing actually 12:46:04 <danshearer1> BKJ621, don't forget (since you are driving) that Ruben says he knows the free parking places 12:46:26 <danshearer1> Final call... are we finished with the Brussels topic? 12:46:33 <Labhraich> Think so. 12:46:46 <Labhraich> I'm not going to volunteer for a 3rd talk unless there's really a lot of beer 12:46:49 <BKJ621> I am 12:47:22 <danshearer1> #topic Next steps in docs/benchmarking 12:47:45 <danshearer1> #info We have increased the combined dataset a bit, but more is needed 12:47:59 <danshearer1> #action danshearer1 to submit a large cluster run asap 12:48:09 <Labhraich> I think the benchmarking framework is stable now - at some point we'll need more tests to make it more representative 12:48:21 <Labhraich> But for now, just run as many as we can 12:48:34 <danshearer1> #info Gabby has got an R Shiny server working. 12:49:31 <Labhraich> For entertainment value - I'm going to run a set on a 32-bit Atom of the most ancient type. May take a long time... 12:50:01 <danshearer1> #action gabby_bch to finish graphs to the point where we can compare performance of all SQLite vs SQLite, and all LMDB vs LMDB, in addition to what we currently have 12:50:16 <gabby_bch> I was thinking that the runs which don't have disk and cpu identification should prpbably be redone since we can't comapre them to the others in a sensible way 12:51:02 <Labhraich> We can add this information if we still know where they ran 12:51:18 <gabby_bch> do we know? 12:51:36 <Labhraich> FOr most of them, yes 12:52:39 <danshearer1> For example, anything with dan-skylake ran on the Rotterdam server in a VM with a spinning disk underneath it. 12:52:49 <gabby_bch> #action gabby_bch fill in the gaps in the combined bechmark database 12:54:41 <Labhraich> Anything from last year is likely to be produced by me, and I can find out where I ran it 12:54:42 <danshearer1> So that would mean we can't re-create the combined database from the individual ones. Does that matter? 12:55:28 <Labhraich> I would probably prefer updating the individual databases, maybe moving them out of the way and using benchmark-filter to make a new copy while adding the extra information (which has options to do) 12:55:34 <danshearer1> BKJ621, we are referring to the databases at https://lumosql.org/dist/benchmarks-to-date/ 12:56:53 <danshearer1> Ok so the originals can become aZwhatever at the bottom, probably one day to be deleted. 12:57:42 <Labhraich> I'm going to get a copy of all (excelt zZ*) and see what has and what doesn't have CPU and DISK comments 12:57:53 <Labhraich> Then see if I can figure out what would need to be added 12:58:29 <danshearer1> #action Labhraich to investigate next step for adding CPU and DISK to older benchmarks, Dan providing data for his runs 12:58:39 <danshearer1> ok good. 12:58:43 <danshearer1> So here is my plan 12:59:14 <danshearer1> As soon as we have cleaner data and a nicer graph, I will update the benchmark discussion document 0.6 and it will become a public document 13:00:08 <danshearer1> #action danshearer1 to incorporate better data, more data and better graphs into a new version of the benchmarking release doc, which can then be committed to the repo 13:00:48 <danshearer1> which is exactly what we want to happen before giving talks in Brussels next month 13:01:33 <danshearer1> have we finished on this topic? 13:02:20 <danshearer1> Looks like we have 13:02:38 <danshearer1> then just pretty quickly... 13:03:05 <danshearer1> #topic LumoSQL internals and next steps 13:03:36 <danshearer1> I'll give a quick summary of where I think we got this morning (and Labhraich can argue if he wants) 13:04:34 <BKJ621> Sorry I am a bit distracted. Seems there is a majority in Sweden & Finland for Nato. The real thing 13:04:52 <BKJ621> Not th option 13:05:08 <danshearer1> #info Labhraich is continuing to work on a general mechanism for adding data to SQLite databases so that unmodified SQLite does not notice 13:06:15 <danshearer1> #info This can be seen as adding internal structure/data models to rows and tables that has never existed before. 13:07:47 <danshearer1> #info Labhraich wants to be able to store metadata about blobs in SQLite databases 13:08:01 <danshearer1> #info Ruben is working on ways to implement blobs that have different levels of data access and control, as well as integrity and uniqueness 13:08:57 <danshearer1> #info Ruben is about Lumions, whereas Claudio is about implementing either Lumions or some other scheme 13:09:30 <danshearer1> #info Dan has received helpful feedback about the LMDB backend and alternative source trees, and is following that up. This was in response to the Benchmarking 0.6 document. 13:10:28 <danshearer1> #action Dan to poke the people who recieved 0.6 as soon as we have slightly better data that shows more obviously the ups and downs in SQLite and LMDB. 13:10:40 <danshearer1> BKJ621, wow. 13:11:07 <Labhraich> Does somebody feel threatened by a KGB criminal? 13:12:50 <Labhraich> So - I see 8 files without DISK and/or CPU comment, 4 of which are mine and I can find out where I ran them, 3 are skylakeibrs and 1 is lumosql-all-lmdb-versions-sqlite-3-34-0-0-2022-m02-07.sqlite 13:13:02 <danshearer1> On a related note, then, and also aware that these meeting minutes are about to become a public document, I can say that the project you have all heard of to modify Signal Server source code was approved yesterday. 13:13:45 <danshearer1> It is easy to imagine how it might be nice to start up your own little Signal server for a group of people right now if you lived, for example, in .ua 13:16:37 <danshearer1> I can't remember what lumosql-all-lmdb-versions-sqlite-3-34-0-0-2022-m02-07.sqlite is. Did I generate it? I think I did. It's laptop SSD if so. 13:16:45 <danshearer1> And Intel core i7 13:16:59 <Labhraich> Might have been on the cluster? 13:17:23 <danshearer1> ahhh - yes it was. Hence the slightly unusual name because Ruben wasn't so familiar with things. 13:18:05 <danshearer1> gabby_bch, looks like we can fairly quickly fix this data for you 13:18:25 <Labhraich> I'll update my 4 and send you the command to repeat the same update on the rest when you have the information 13:18:33 <danshearer1> Labhraich, ack 13:19:09 <danshearer1> Ok that's all I have on this topic. I would not be surprised if there are some recommendations for LMDB tuning coming from various discussions. 13:19:41 <danshearer1> does anyone else have any other business? 13:19:46 <danshearer1> or pleasure? 13:20:18 <BKJ621> not me 13:21:26 <danshearer1> ok I think we are probably done 13:21:31 <danshearer1> and thankyou all 13:21:52 <BKJ621> Same to you!! 13:22:36 <danshearer1> waves to Wales, France and Sweden 13:22:51 <danshearer1> #endmeeting