 The T2 Tile Project is building an indefinitely scalable computational stack. Follow our progress here on T Tuesday Updates. Happy New Year. Top Stories this week. Working on the novel. It's just not working. Thinking I can go back and forth and do a day or an afternoon on writing. So like I said last week I'm gonna stop talking about it for a while and I'm gonna push it off until March after the deadlines that I'll talk about for that are coming up for the Artificial Life Conference. I'm gonna dedicate a week to try to keep moving the ball so we'll see how that goes but won't talk about it between now and then. Finally, finally we've got Debian 10 Buster running on the T2 tiles. We've had it mostly working last week except we couldn't get the touchscreen to work which had been a terrible problem previously but it seemed to be a different problem and so that means we have leaped all the way from 4.4 Linux kernel 4.4 as of 2017 to Linux kernel 4.19 as of 2019 and eventually you know I was tearing out what little hair I have left trying to figure out what had changed and so eventually I had them running side by side I mean so this this gray window is the old Keymaster running 4.4 this dark black window is running 4.19 it's the red Keymaster you know where are they different where are they different and you know the problem I had getting touchscreen going in 4.4 was a hardware problem a hardware configuration problem that I had wrong but that wasn't it and so in particular looking at the newer stuff I was drilling down into these incredibly deeply buried things where you can get at information about what the communications hardware is doing and in particular the spy interface this ADS 7846 that's the chip that the touch screens use the screens that we've got used to translate actual touches on the screen into numbers that get sent to the to the Linux box and you know the bytes received were going up you know I would touch the screen and and move it around and there would be stuff received I would check the number of interrupts that there was supposed to be an interrupt every time that pen finger whatever goes down to when it goes up and the interrupts were going up so it wasn't that but I couldn't for the life of me get any events to get delivered and once I started in figuring out how to get events to get delivered they were all messed up and you know in the end right so so there you know cat proc interrupts I could search through the list of all the interrupts and find ADS 7846 and say yes indeed touch the screen interrupts go up finally the answer was that the SDL package the simple direct library something that the MFM T2 code uses to display everything the atoms and what not and to handle input mouse touches touch screen touches and so forth was busted as far as this particular touch screen worked and so the solution which I have now implemented is we build the SDL ourselves we make a dummy package so that the overall package management system doesn't get upset by the fact that we're not installing the official one that's supposed to be there in particular all the other packages like having to do with fonts and so forth we don't have to actually build them long and short of it is with you know the month of November taken up trying to write a novel the best effort the month of December taken up trying to upgrade Linux and finally finally here we are in January in a new decade and we have just gotten this actually working so we are sort of back where we were two months ago as far as overall functionality of the T2 tiles and still are facing getting intertile events working but there was one more issue I yeah so the red key master was going to be the replacement for the white key master that had been using all along back in the four four year to actually have the private signing key on it so that new versions of the software could be released that would be accepted by the other tiles that have the public key for the common data manager compiled into them so that we could spread stuff out and the red key master was going to be the new version of it with red because I had printed up a nice red case using this PETG stuff of that I really like as a replacement filament for the PLA that I had done everything with for most of the project and you know so here you know every day if you want to when you need to reflash the whole disc on a T2 tile you need to take the cover off so that you can get to the slot where the mini SD card micro SD card goes in so this thing I was taking on and off so much that I just put rubber bands on it to hold it in I hadn't even screwed it down over the last several days when I finally got Debbie in 10 all working because I started building SDL myself I wasn't actually using this guy because I was out and I just building it on another tile because you know it doesn't really matter until a signing key actually gets put on them and so the actual last guy to get flashed with the new software was the red key master I screwed him into I pet plugged him into the power zone that I've been building and rebuilding and rebuilding as I go through all this and this touchscreen didn't work on the red key master the touchscreen worked on the other 15 tiles in the power zone not on the red key master I thought it well you know maybe it's this the screen had gone bad so I replaced it with another one and that didn't work either so then I felt bad and I thought well maybe it's because I had failed to put insulation between the back of the display which is what you're looking at here and the top of the Beaglebone green and maybe they were shorting out somehow so I took off the screen cover the you know the protection film and I put it on the back to cover up to insulate the thing it didn't help so I swapped it out entirely for a whole different display I got that proper capped on tape and set that up so that you know yeah it's it's insulated you so I mean you can see there the the exposed pins on the back of the Beaglebone green are right there and you know so maybe that was it no it wasn't and I put the whole thing back together again and as I was putting it back together again it started to seem like if I actually screwed the screws down snug not even super tight you already have to kind of leave the southwest case screw a little bit loose or else the power switch kind of gets jammed down because the tolerances are a little tight but I knew about that and that's okay this is it more in the northwest corner although it seemed to be kind of in any of them and what I found was if I screwed it down not even that tight that whoops yeah that so the interrupts were starting to go crazy so I wasn't doing anything I was just looking at this every two seconds the thing was updating and the number of interrupts were going crazy like it was seeing a permanent hit so it was going I got something got something got something and in fact it was never seeing up it was never seeing the release and so it wasn't actually doing anything finally I backed off the screw a little bit at seventeen thousand two hundred interrupts and then the interrupt stopped so as far as I can tell and once I backed off and got it loose enough then the touchscreen started working I could bring up the command menu and cancel it and do things and so forth so as far as I can tell the mystery of the red key master is the red key master case I switched it out for a PLA one of the black cases that I just had sitting around and it's fine I can screw the screws down to the same amount of tension that I expected to and everything works fine so at the moment the red key master is black doing this read about it as I have to read tiny little red buttons on the whole thing and that's it so that's a bit distressing I don't know if there's something I mean if PETG prints slightly thicker or shrink or something I don't know but the red key master case seems to be systematically associated with the touchscreen not working which is a little bit scary just the mechanics are that tight but I've not seen this with any of the other ones that I've know that I've dealt with you know dozens anyway so I don't know so we'll see and so here's the red key master and you know it looks like it's got some red at the bottom well that's just the speed frame for putting the nuts in and out that's not actually the thing but then again with the new case on it it all starts working fine and and this was a fun one I had the camera actually caught it halfway between the display of dismissing the menu and going back to the base one so yeah and then the interrupts act reasonably you know when you touch they go up when you don't they don't so that's it what I wanted to try to do here is let's do a live demo okay so I've got the whole power zone here let's power it up and see if we really can get 16 out of 16 tiles to come up it still takes an awful long time for them to come up takes about 80 seconds for them to come up which is sad I'll be able to speed it up a little bit but the main problem seems to be even though the shutdown all seems to go correctly it seems like every time it comes up it wants to file system check the disk and it burns up like 25 seconds doing that that it's really not supposed to need if it's shut down cleanly so I don't understand that and then there's a bunch of other little stuff that's still happening because I use the internet of things distribution that has so much stuff stuffed into it so that it might be possible to speed up the boot a little bit going forward in the make things better principle than we've gotten here so far but what I'm really concerned about is everybody look at that look at that everybody came up that makes me so happy after all this mysterious system d failures now it takes a very long time for all of the MFM to actually get started running and officially I want to count the boot time up until MFM is doing events but the fact that these things take so long is mostly because they're actually getting wedged in the fact that the intertile stuff is not working yet so these these things have started to see things see stuff coming from the neighbors which has caused them to get lock lock locked up which is causing them to do time outs and so forth so hopefully these guys will all come yeah so they're starting to so that's there we go everybody is up they're all pretty well wedged because we don't have an entire working properly yet but the touchscreen is working we can do grid-wide events like get rid of the statistics screen look at that get it back see now did we not see some kind of the grid the grid prefix sometimes doesn't quite work the way I expect it to so I think that's that that sent a command around to everybody to turn it off so well let's try somebody else grid what the heck shut down and again some of these guys are in the process of timing out from being wedged about intertile stuff being messed up so hopefully when those guys finish timing out they will go ahead and process the the grid timeout message maybe we shouldn't stay because we have other stuff to talk about but oh there they go okay now I feel like we've actually got a reasonable base to put our work on intertile events on top of so all right finally just at the end here it's now seven weeks to March 1 as far as t Tuesday updates are concerned which is the deadline for submitting the scientific paper that's not the only deadline of this conference that I want to submit to the art robotics displays and visualizations which is where the t2 tile grid is going to get submitted as an art robotics display and visualization that's due May 15th that's later so that's okay but there's also workshops tutorials and special sessions and I think you know really it's time to do an oolong splat introductory tutorial so I think I'm gonna try to write a tutorial proposal and that's due January 17th I thought it was due February 17th oh my god has that changed I'm not sure if that's changed I thought it was February yeah so well so that's really quick yeah and we'll see it this the tours are only an hour and a half long which is incredibly fast but we'll generate a some lecture out of it and we'll get video and there'll be tutorial exercises and stuff that will be hopefully a value going forward even after the tutorial regardless of whether anybody actually shows up I mean hopefully somebody will show up and we'll even get as far as maybe people be able to do some simple stuff that'll run on the grid by the end of the week of the conference we'll see alright so that's the plan for the next well real soon now and we'll just keep checking off as much as we can get to and then over the next seven weeks happy new year hopefully we're gonna start seeing a lot of progress and a lot of fronts in the next weeks and months 2020 is the year of the T2 tile let's make it so hope to see you next week