Monday, 2022-11-21

*** lkcl <lkcl!> has joined #libre-soc01:36
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC06:35
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc06:36
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:31
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:31
ghostmansd[m]lkcl, so, if there's nothing wrong, we all should submit RFPs, right? :-)07:50
ghostmansd[m]Could you send the top-secret link, please?07:50
*** octavius <octavius!> has joined #libre-soc08:57
*** toshywoshy <toshywoshy!~toshywosh@2a02:181f:0:1e9:41c6:f2e2:2466:f85b> has quit IRC09:14
*** toshywoshy <toshywoshy!> has joined #libre-soc09:14
*** openpowerbot_ <openpowerbot_!> has quit IRC09:15
*** openpowerbot_ <openpowerbot_!> has joined #libre-soc09:15
*** openpowerbot_ <openpowerbot_!> has quit IRC09:31
*** toshywoshy <toshywoshy!> has quit IRC09:31
*** toshywoshy <toshywoshy!> has joined #libre-soc09:32
*** klys <klys!~mdasoh@> has quit IRC09:51
*** klys <klys!> has joined #libre-soc09:51
toshywoshyam I not awake enough or is Libre-SOC missing from the NGI pointer overview10:19
octaviusWhat an exciting event :)11:38
markoswhat event?11:58
octaviusngi pointer had a presentation this morning. Apologies, not sure if it was public or not12:23
octaviusAll the projects are open-source though, was interesting to hear their progress12:25
markosah ok, no need to apologize, not sure I would be able to watch it anyway, lack of time, etc12:28
markosbut nice to know there was one12:29
markosbut why was LibreSOC missing?12:29
octaviusToshaan meant that Libre-SOC didn't give a presentation. There were already 6 projects represented (and a few more described in chat). Time ran by pretty quick too.12:38
toshywoshyno, I meant I did not see our logo on the list12:53
octaviusAh sorry, thought you meant presentation13:18
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC15:28
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc15:29
*** octavius <octavius!> has quit IRC16:39
lkclghostmansd[m], the next phase is for the budget-sync program to be run to create the JSON file that needs to be emailed directly to Michiel and the team at NLnet16:48
lkclthat JSON file contains the "back-end format" of the milestones that are DIRECTLY listed in the NLnet RFP system16:48
lkclwithout that (complete, full) list of tasks pre-loaded into the NLnet RFP web server database, you as an MoU Signatory will be presented with NOTHING.16:49
ghostmansd[m]lkcl, do you mean 967..968 tasks?16:50
lkclmarkos, you were not a signatory to the NGI POINTER Sub-Grant Agreement therefore you were not invited to the (private) event16:50
lkcloctavius was not aware of this16:50
lkclghostmansd[m], no, the JSON file has to contain *all* the tasks - and sub-tasks - for the entire #589 MoU.16:50
lkclthe JSON files are presently created but they do not contain the "comments" (needed by NLnet)16:52
lkclif you want to have a look at what the JSON files should contain, run this:16:52
lkclnohup budget-sync -c budget-sync-config.toml -o mdwn --comments16:52
lkclthen have a look at the contents of the file named:16:52
lkclthat is the file that needs emailing to michiel16:53
ghostmansd[m]"Have a look at the contents" — I assume this needs careful inspection. What parts of it should I consider before even bothering Michiel?16:57
ghostmansd[m]What should be the subject? Who, besides you and Michiel, should be added?16:57
ghostmansd[m]This would have been great if we had docs on this besides this IRC chat.16:57
ghostmansd[m]Luke, you want me to participate in management activities. This is OK, but I don't want to fuck the things up. My point is, it'd be great if we have some explicit and documented way to make it work.16:59
ghostmansd[m]I still don't even understand why 966..968 are OK, despite that my nohup.out had some bitching claims about the budgets.17:00
ghostmansd[m]And, obviously, this clearly shows that I will do something wrong, because I don't understand how this stuff works.17:01
markoslkcl, it's ok I understand18:11
markoslkcl, ghostmansd[m] I think we should do a brainstorming session with the sole goal to split tasks/budgets19:00
markosI also agree that it's risky for us to undertake budgeting and administrative tasks19:01
markoswe can't just put numbers in bugzilla and hope they "work"19:02
markosfor that matter we have to do time estimates for each task19:02
markosbecause the previous ones were lacking in that to put it mildly, at least the media codec ones, as it turned out, I'll explain myself: mp3 dct was 750eur, but av1 was 3k, but in fact the mp3 dct was/is *significantly* more complicated to implement rather than the function chosen for av119:06
markosnot only that, but many instructions were missing even and it became possible to write these functions only late in the game19:07
markosI understant that a lot of this knowledge was not available at that time, but now we know better19:08
lkclghostmansd[m], ah - i know what's happened: did you *delete* the nohup.out file before running the command?19:45
lkclbecause nohup performs an *append* to the file, it is not like the "script" command which *truncates* the file named "typescript".19:46
lkclso you are likely seeing an old (former) run, not realising it is "appended" with the newer (completely ok) results, looking *only* at the first line and going "huhn??"19:46
lkclok i'm looking at the online version, here19:48
lkclNLnet.2021.02A.052.CryptoRouter 58919:48
lkclbug #  589  budget  50000 excltasks      0  s,p  130019:48
lkclthat's very very sparse.  only a fraction is allocated to sub-tasks19:48
lkclwe will have to submit this JSON file once19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclthen submit it again when new sub-tasks are added19:49
lkclyou get the idea19:49
lkcli have to talk to Michiel about this19:49
lkclbecause clearly it's not really going to work19:49
lkclbut right now it is the best we have got19:50
lkcli will email you privately the email that needs to be cc'd19:51
lkclmarkos, i've been doing this for 2+ years perfectly fine!19:51
lkclit is a heuristic that has many methods of "compensating for"19:52
lkclsuch as putting a much larger budget for a lot less work to someone for a completely different task19:52
lkclor even giving them some money for a task that they did not actually do (because this is all donations anyway)19:52
lkclbut yes, the Video Grant was the one that was designed the absolute worst, by way of an experiment19:53
lkcland also it was designed with lauri as the main contributor in mind19:53
lkcland, mp3 was literally the first we tried, when there weren't even any FP operations in ISACaller!19:54
lkcli had to scramble non-stop for several days to get the instructions in, in real-time, as lauri was trying to use them19:55
lkclit is similar to what you experienced, with wanting to do DCT in integer (no integer butterfly instruction) but far more pressured because there was *nothing*19:56
lkclnot even any infrastructure to *make* FP operations in ISACaller.19:56
lkclfrom that experience, the lesson is learned, really, that the person writing the assembler really has to be the one that "down tools" on that task and actually adds the missing instruction(s) *themselves*19:58
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC21:26
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc21:27
markoswell, we did what we could with the time we had available, but now we know better, so we should plan better22:09

Generated by 2.17.1 by Marius Gedminas - find it at!