**** BEGIN LOGGING AT Fri May 29 02:59:57 2020 May 29 07:46:11 hello zmatt . I am using a pocketbeagle and trying to use the config-pin utility to configure pins and get info about the pin state but i am getting an errror . I tried "config-pin P2_08 in" and got error " ERROR: write() to /sys/devices/platform/ocp/ocp:P2_08_pinmux/state failed, No such device". I even tried as root May 29 07:48:01 I also checked the P2_08 comes out to be gpio60 and i can see a entry exported in /sys/class/gpio May 29 08:00:02 hello again, I am following this tutoriel on how to use beaglebone PRU May 29 08:00:21 https://pastebin.com/pBpMZPGj May 29 08:01:43 and the last command returns "-bash: echo: write error: Invalid argument" May 29 08:01:53 i want to know what'ss causing this error May 29 08:03:45 the kern.log file contains this May 29 08:03:52 https://pastebin.com/d602xU3Q May 29 08:32:27 i want to know is the pru cape necessary if i want to use the PRU ? May 29 09:24:26 (no, but having the patience to wait for a reply might be necessary :P ) May 29 09:31:51 interesting, that error means a .resource_table section was found that's smaller than the size of a resource table header (16 bytes) May 29 09:31:58 how'd he manage that May 29 09:42:44 deepankarmaithan: the error "No such device" (ENODEV) coming from config-pin (or more generally writing a pinmux-helper's state attribute) indicates you tried to switch it to a non-existent mode May 29 09:46:10 deepankarmaithan: P2.08 only supports the gpio modes (gpio, gpio_pd, gpio_pu) May 29 09:47:10 (the only other mux mode it has that's of any use is mcasp 0 rx clk, but you can't select that with config-pin since using a mcasp requires DT configuratoin anyway) May 29 10:59:09 sorry i got disconnected May 29 10:59:38 do i have to use a pru cape in order to use the PRU on the BeagleBone Black May 29 10:59:55 no May 29 10:59:56 the PRU is part of the am3358 May 29 11:00:31 exactly... the PRUs are just two specialized cpu cores that are part of the SoC May 29 11:01:28 what about implementing the .out file to the pru core from the code composer studio, is there a way to cheat using emulators ? May 29 11:01:35 avoid* May 29 11:01:50 the pru cape just seems to be intended as a way to demonstrate some of pru's abilities May 29 11:02:02 I have no idea what you mean May 29 11:03:01 i want to run a c code written in the TI code composer studio on the pru May 29 11:04:19 that requires a jtag connection I think, although I'm not sure why... in princple it could just use some sort of remote debugging stub May 29 11:06:40 i was trying to follow the TI training series about PRU and I wanted to confirm that i don't need a pru cape and if i can find another way to implement the .out file to be executed by the pru in another way May 29 11:07:06 i tried remoteproc but i am stuck in an error May 29 11:07:17 following this exemple May 29 11:07:43 https://pastebin.com/p9n23pxK May 29 11:09:05 why are you giving a link to a pastebin that just contains a link? instead of just giving that link directly May 29 11:09:31 and yeah, that error you got ("header-less resource table") is a pretty weird one May 29 11:09:40 sorry i thought it would be wrong to send the link directly May 29 11:10:21 use pastebin for stuff that's multple lines May 29 11:11:36 anyway, "header-less resource table" means there's something wrong with the binary: its resource table is smaller than the 16-byte header it's supposed to contain (at minimum) May 29 11:13:03 okay i'll try building another version of the .c file to get another .out file and retry* May 29 11:14:04 can you do readelf -S your-program.out | grep resource_table May 29 11:15:26 i got the following result May 29 11:15:28 [14] .resource_table NOBITS 00000000 000000 000000 00 0 0 1 May 29 11:15:33 yeah that's not good May 29 11:16:35 that means you're missing the resource table May 29 11:16:48 can you please further explain why ( i am new to pru and beaglebone in general and i want to learn ) May 29 11:16:58 this isn't a PRU-thing, it's a remoteproc-thing May 29 11:17:47 the resource table is just a datastructure for special stuff that needs to be setup by the kernel for you. even if you don't need any of that, you still need a resource table May 29 11:17:59 https://pastebin.com/G3Xt3uPi . Please look at this link. Even config-pin -i switch is giving a error May 29 11:18:18 that's why many examples have #include "resource_table_empty.h" May 29 11:18:39 which defines an empty resource table in the correct way May 29 11:19:32 ah okay i didn't include a resource_table.h in my .c file i'll rebuild it and try again May 29 11:19:36 thank you May 29 11:20:28 deepankarmaithan: config-pin has no -i option, so it's saying to configure the pin named "-i" to mode "p2.08" May 29 11:20:33 *it's trying to May 29 11:21:28 and there's no pin named "-i" May 29 11:21:29 :P May 29 11:21:34 config-pin is just a bit stupid May 29 11:24:43 hmmm . Exploring beaglebone book says it does support -i but the config-pin help tells me that it doesnot May 29 11:51:15 how old is that book? May 29 11:51:43 ohey, config-pin also isn't a shell script anymore May 29 12:41:28 m May 29 12:46:56 the am33xx doesn't have a temp sensor, right? May 29 12:57:33 TRM says the ADC has an accessible temp register. Nothing directly temp-related like the AI's uC has. May 29 15:53:21 set --> https://pastebin.com/HY5rEzwF May 29 15:53:29 err set_ --> https://pastebin.com/HY5rEzwF May 29 16:16:38 mru: it does yeah, a bandgap sensor, but it is so shit that it has been deemed useless, so TI defeatured it May 29 16:17:17 is it too shit to even tell if the chip is running relatively hot? May 29 16:17:53 if you google "am335x bandgap" you'll find the relevant E2E thread May 29 16:18:05 yes, I already found some discussion May 29 16:19:19 I was thinking it could be useful to monitor the temperature so if a device fails we could check if it has been hotter than others May 29 16:19:23 ±10.8 ͏°C apparently May 29 16:19:40 that is indeed a bit shit May 29 16:19:41 oh, comparing between devices sounds even more iffy May 29 16:20:21 noticing an increase for a single device might be useful too May 29 16:22:00 it's not important May 29 16:22:01 Yes band gap devices require a single point calibration. TI does this on other devices for you, but likely not on the AM335x May 29 16:22:07 these devices have been pretty reliable so far May 29 16:23:54 it would be useful to notice large delta's probably? May 29 18:22:16 So I'm trying to use my new kernel and it doesn't have any modules. I mean it has them all, but they're all .xz and don't appear to be recognized as modules. Searching so far has yielded no intel. What simple thing am I missing *this* time? May 29 18:24:00 xz support is probably missing in the kernel or kmod May 29 18:24:54 Why would it generate xz if it doesn't have support? May 29 18:25:25 does modinfo work? May 29 18:25:37 Dunno. Lemme check. May 29 18:26:58 Odd. Says "not found" when I'm in that dir and tab-complete works on it. It's not zero len and I have permissions. May 29 18:27:58 I'm tempted to just extract them all to .ko I know is in there and not worry about it. May 29 18:29:50 seems like kmod is built without xz support May 29 18:59:30 So I did extract them all. Now depmod -v shows pretty much every one of them lacks a dependency, where before it output nothing. I guess that's progress? May 29 18:59:59 at least it's recognising them as modules now May 29 19:00:33 what's the missing dependency? May 29 19:01:22 Gobs and gobs of different ones, like "usb_interface_id", which strikes me as a bit odd. May 29 19:01:43 It's like it doesn't know any libs exist. May 29 19:02:01 is it using the wrong kernel somehow? May 29 19:02:16 Going to look at that, now that I've typed it. Maybe this one doesn't have that set right either. May 29 19:02:30 uname -r is right, if that's a metric. May 29 19:02:57 I've never fiddled with replacing the kernel so I'm sure I'm doing something bonehead. May 29 19:03:25 we've all been there May 29 19:03:35 Sure. Now it's my turn! \o/ May 29 19:03:49 usb_interface_id is defined in drivers/usb/gadget/composite.c May 29 19:04:24 and that becomes part of the libcomposite module May 29 19:05:09 do you have that modules? May 29 19:05:15 -s May 29 19:05:47 Doing a find. May 29 19:07:26 Says it's under drivers/usb/gadget. Huh. It even lists libcomposite.ko in the message. May 29 19:07:46 Maybe my module expors are messed up? May 29 19:08:07 what exactly does the error look like? May 29 19:10:00 This is one. I have 1771 of them total, all manner of needs for all manner of modules. May 29 19:10:09 lib/modules/4.19.106-bone49/kernel/drivers/usb/gadget/function/usb_f_acm.ko needs "usb_interface_id": /lib/modules/4.19.106-bone49/kernel/drivers/usb/gadget/libcomposite.ko May 29 19:10:52 are you sure that's an error and not merely a statement of fact? May 29 19:11:27 oh, you ran depmod -v May 29 19:11:34 the -v flag does exactly that May 29 19:14:33 Ok. So those aren't errors, it's the dep list. Then why is modules.dep empty? May 29 19:19:56 no idea May 29 19:20:04 don't recall ever having had that problem May 29 19:20:40 Clever. It's gui madness. When I look at modules.dep in CLI it's populated, and when I reboot I have modules now. \o/ May 29 19:21:01 how were you looking at it when it appeared empty? May 29 19:23:04 Remote dolphin session from a Linux system, where Linux is access via X, so it's "remote" as well. I should have know to check CLI ... it's bitten me before. Works a treat for drag/drop, kinda wonky otherwise. May 29 19:24:07 I cross-compile on Linux and drop the files onto the B'bone. I could probably automate that, but this works easily enough. May 29 19:24:33 rsync is nice for sending files to other systems May 29 19:25:34 Yup. I know. I've just never done that for whatever reason. (shrug) Perhaps if I did this more frequently. May 29 19:25:51 I only get to play with this about once a year. (pout) May 29 19:26:12 https://xkcd.com/1319/ May 29 19:27:00 Total sucker for xkcd! lol May 29 20:22:47 Ragnorok: you're not using a kernel .deb? May 29 20:23:09 That's how I got this onto the system, yes. May 29 20:23:49 I used the .deb I built yesterday. May 29 20:23:58 installing a custom kernel is normally as easy as: use rcn's (re)build_deb.sh script, copy the linux-image (and optionally others like linux-headers) package into the target, install with dpkg May 29 20:24:23 And I did those things today. But the modules were all compressed so it wouldn't load anything. May 29 20:25:08 I mean, that depends on how you configured the kernel May 29 20:25:45 CONFIG_MODULE_COMPRESS_XZ May 29 20:25:53 I left it at whatever default it had on the premise rcn's scripts were right. May 29 20:26:36 I may have "left" it wrong. When menuconfig came up I did "exit" and let it drive on. May 29 20:26:53 yes, rnc's kernels use it: May 29 20:27:30 /lib/modules/$(uname -r) on the latest IoT image contains only .ko.xz files, no .ko May 29 20:27:33 it works fine May 29 20:27:40 the kernel decompresses them on the fly May 29 20:27:45 I also figured if it compressed the modules it would load them, but it did not. May 29 20:28:14 I ran unxz on them all and they work now. May 29 20:28:38 what userspace are you using? May 29 20:28:52 what version of kmod May 29 20:30:02 kmod --version on current buster images show it's been compiled with XZ support May 29 20:30:04 It says 18. May 29 20:30:09 (version 26) May 29 20:30:56 in fact I'm assuming "+XZ -ZLIB +OPENSSL -EXPERIMENTAL" means the only compressed format it supports is .xz since support for .gz has been disabled May 29 20:31:21 I'm also guessing your kmod doesn't have +XZ :P May 29 20:31:47 Prolly not; mru said as much earlier so I uncompressed them all. May 29 20:31:49 at least that would explain why it didn't understand .ko.xz modules May 29 20:32:07 Mabye I shoud update that? May 29 20:32:56 The reason I built this kernel is so I'm running something maybe new enough I could get help with a USB issue I'm having, where FAT32 reports a flash drive is full when it's clearly not full, nor is the directory, nor does any available data indicate why it's saying the drive is full. May 29 20:33:14 either make sure your kmod supports xz-compressed modules or disable CONFIG_MODULE_COMPRESS ("Compress modules on installation") in your kernel config May 29 20:33:30 Okey doke. May 29 20:35:51 If there's some other option I could explore I'd be happy to do so, with the understanding I've no xp at this level of Linux, but I'm fine mucking around with whatever. May 29 20:36:57 how is it reporting it is full? May 29 20:37:45 what does df say on it? May 29 20:37:55 Latest bit was a cp says cannot create regular file no space left on device. May 29 20:38:02 DF says it's half full. May 29 20:38:21 Around 4GB on an 8GB stick. May 29 20:39:14 My recollection is I researched FAT32's limits and I shouldn't be near any of them, but I don't recall the specifics. May 29 20:41:42 Ragnorok: can you run "dosfsck -nv" on your dos filesystem? (preferably in "full" state) May 29 20:42:18 Will shortly. I'm in the middle of copying everything off it on another box. May 29 20:43:11 the file it is failing on - how big is it? May 29 20:43:42 in kernel 4.14.78 there only seem to be two cases where fatfs returns ENOSPC: cluster allocation failed (i.e. truly out of space) or too many entries in a directory May 29 20:43:46 is it anywhere near 25% of the space? May 29 20:43:48 Looks like 126k May 29 20:43:58 what does df -i say? May 29 20:44:11 Have to wait until I can plug it back in. May 29 20:44:16 'k May 29 20:44:37 df -i returns all-zeros for fatfs May 29 20:44:52 grrrrr May 29 20:45:16 seems like filesystem issues come in droves May 29 20:45:30 spent good part of last night figuring ways around HFS May 29 20:45:32 I don't think that's an issue, I don't think fatfs has inodes May 29 20:45:43 I don't either. May 29 20:45:54 I'd think dirent would be mapped to inodes May 29 20:46:00 oh are you copying to the root of the drive? May 29 20:46:09 No. May 29 20:46:13 fat has a limit on number entries on the root dir May 29 20:46:17 Yup. May 29 20:46:27 on every dir in fact :P May 29 20:46:33 I think? May 29 20:46:38 df -i is all zeros. lol May 29 20:46:40 just rootdir May 29 20:46:51 does dmesg show any more details? May 29 20:48:02 dosfsck is in progress. May 29 20:48:08 well, at least linux fatfs has a max number of entries per directory for every directory: https://elixir.bootlin.com/linux/latest/source/include/uapi/linux/msdos_fs.h#L34 May 29 20:48:23 line 33 sorry May 29 20:48:26 'k back to insane hand soldering of 0.4mm connectors :( May 29 20:48:41 fun May 29 20:48:45 I'm not at 65k files. May 29 20:49:00 don't understand dev boards using 0.4mm connectors May 29 20:49:18 esp for signals like UART and GPIO May 29 20:49:25 dmesg doesn't seem to say anything. I shows the plugs/unplugs. May 29 20:49:38 odd May 29 20:49:44 are you sure there is that much space? May 29 20:49:58 i.e. you didn't dd in a blank FS from a bigger device? May 29 20:50:02 0.4 mm is for qfn, annoying for connectors May 29 20:50:09 unless high-speed stuff May 29 20:50:19 exactly May 29 20:50:37 but that's probally why this dev board is $20 cheaper then the other unofficial ones ;) May 29 20:50:39 dosfsck says 488k clusters and 248k used. May 29 20:51:05 I have dd'd to this stick at all, no. May 29 20:51:35 /dev/sda1: 28733 files, 248359/488304 clusters May 29 20:51:48 that is a lot of files May 29 20:52:13 But it's not 65k in any single directory. This is a lot of small files. May 29 20:52:40 First thing I thought is I'm hitting some file count limit. May 29 20:55:32 It IS a dir limit! If I copy to a different dir it works. May 29 20:55:45 Ragnorok: I'd just put some dynamic debug lines (dev_dbg() or pr_debug()) on every ENOSPC condition in fs/fat/ May 29 20:55:55 ah May 29 20:55:56 never mind May 29 20:56:22 I'm a bit confused why it's a dir limit though. May 29 20:56:40 though, if you're not having 64K it's still a bit odd May 29 20:56:55 actually, I don't know how many directory entries are used by a single file May 29 20:57:01 I'm nowhere near that. I'm not even half that, which would have been something. May 29 20:57:27 Ah. Is there a way to tell how many dir entries are in use? May 29 20:59:09 at least, I vaguely recall FAT using multiple directory entries for stuff like long filenames and extended attributes, but maybe I misremember May 29 21:00:05 "A long file name for a target file or sub-directory is stored in a set (one or more) of additional directory entries associated with the short name directory entry describing the target file or subdirectory." -- FAT spec May 29 21:00:19 lol May 29 21:01:19 Whelp more next week. Have a great weekend all. As always, thanks for your time and suggestions! May 29 21:01:43 so yeah, that means you can hit that 64K limit without having 64K actual files in a dir May 29 21:02:09 Yeah. Monday I'll see if I can discover how to tell what's in use. May 29 21:02:31 I need a way to know this might be happening. Enjoy! May 29 21:02:40 long filenames May 29 21:02:45 I thought the root dir was file number limited by the sub directories weren't. May 29 21:02:50 11.3 = 1 entry per file May 29 21:03:41 8,3 sorry May 29 21:04:34 the long name uses one directory entry per 13 chars May 29 21:05:09 really_long_filenames_could_make_a_serious_problem_then.txt? May 29 21:05:19 GenTooMan: no, this applies to all directories May 29 21:05:38 and it looks like maybe the root directory has an additional limitation for FAT12/FAT16 but not FAT32 ? May 29 21:05:43 not sure May 29 21:06:53 I say this based on this code: https://elixir.bootlin.com/linux/latest/source/fs/fat/dir.c#L1297 May 29 21:07:12 hmm I think the root dir has a limit on FAT32 as well. It's the sub dirs that aren't (as) limited? The easy way is to check FATFS. Still using a direntry for each 11 characters is a bit silly. May 29 21:07:40 this seems to imply that the root directory can't grow May 29 21:07:44 except on FAT32 May 29 21:08:00 but FAT_MAX_DIR_SIZE applies to all directories May 29 21:08:39 hmm so 64K entries can become 10k entries with careless use of long file names. May 29 21:11:44 is there a need to share data or actively access the data? if not, tar it up May 29 21:12:27 worse yet, it needs to allocate a contiguous block of entries (to accomodate the long name), so it could also fail to allocate due to fragmentation of free directory entries May 29 21:12:41 ds2: Ragnorok has left already May 29 21:16:36 memories of norton disksort May 29 21:18:30 I don't really understand how that code handles the case where the directory needs to be grown (i.e. fat_get_entry-loop will exhaust all allocated entries and if(nr_slots) will be hit later) yet the new size would be greater than FAT_MAX_DIR_SIZE May 29 21:19:16 could it be subdir's can be chained like a linked list? and that limit is really on per unit? May 29 21:20:17 the limit is on the total. hitting it will not grow the directory, it will fail with ENOSPC May 29 21:20:30 growing the directory happens under the condition I just mentioned May 29 21:21:09 3 May 29 21:22:16 if I actually cared about fatfs I'd do some simple tests to confirm the behaviour, should be simple enough to do May 29 21:23:52 found another source that confirms it. 64K 8.3 entries per dir May 29 21:29:30 oh it's not 13 bytes per long-file-name-entry, it's 13 16-bit "unicode" chars, though it's not clear if they mean utf-16 codeunits or just ignore the existence of codepoints above 0xffff :P May 29 21:31:04 64k files ought to be enough for everyone May 29 21:39:22 ok I tested it anyway: https://pastebin.com/raw/urGc67Ln May 29 21:39:46 so yeah, 3120 files whose name is longer than 247 chars May 29 21:40:05 yields 3120*21=65520 directory entries May 29 21:50:10 still 12 bytes per dir entry consumed is kind of nasty. May 29 21:50:37 although I don't know what the alternative might be... May 29 21:56:47 Well I am sure they had a reason to not use the rest of the dir entry for file name data, I'm just not sure why yet. May 29 22:01:05 compatibility presumably May 29 22:01:16 also, confirmed, fat directory fragmentation is definitely a thing: https://pastebin.com/raw/j95FXcen May 29 22:02:28 if there's a wrong way of doing something, you can be sure microsoft found it and executed it badly May 29 22:02:47 I mean, fat was designed to be super-simple May 29 22:02:54 easy to read and write by really limited software May 29 22:03:05 especially read May 29 22:03:18 but directories were a bit of an afterthought May 29 22:03:29 or did those early dos version not use fat? May 29 22:03:50 dunno, but regardless, the directory structure is fine given the goal of simplicity May 29 22:04:06 and assuming some legacy notions on acceptable file name length ;P May 29 22:06:04 and the fatfs driver could have implemented some code that (partially) defragments the directory if there's not enough unfragmented space... it chose to just keep things simple instead :P May 29 22:09:27 So the long file name system requires contiguous directory entries to work? May 29 22:09:32 yes May 29 22:10:31 the long file name entries immediately precede the real directory entry May 29 22:10:32 at least admit it was idiotic of them to name the place where software gets installed PROGRA~1 May 29 22:10:51 mru: ? why? May 29 22:11:22 why would you even be poking around there with software/tools that doesn't understand long file names :P May 29 22:11:23 why insist on "Program Files" and get the ugly mangled name when "Programs" would have been just as clear in under 8 chars May 29 22:11:35 do you not remember the 90s? May 29 22:11:44 I was a mac user in the 90s May 29 22:11:45 half the time you'd get the mangled name for no apparent reason May 29 22:12:04 hmm looks like they didn't make it simple but more of a quick hack. May 29 22:12:17 I used Linux in the 90's do I count? May 29 22:12:19 GenTooMan: the long filenames are a hack, but that's what you get if you need backwards compatibility May 29 22:12:22 probably depending on which of the two dozen almost equivalent syscalls the app used May 29 22:13:21 syscall being used in the loosest possible sense May 29 22:19:39 mru: note btw that "Programs" would still require a long name (due to not being all-uppercase) and FAT doesn't specify any relationship between long names and short names (hence "Program Files" could have been PROGRAMS and "Programs" could still be PROGRA~1) May 29 22:19:53 well no use beating dead problems to death. However one can at least learn from them. May 29 22:20:42 zmatt: I never saw mixed-case filenames of at most 8 chars have a short name other than the all-uppercase version May 29 22:21:05 sure, that's the sane thing to do, but technically it's an implementation detail :P May 29 22:21:28 a decision made by the OS, not by the FAT spec May 29 22:21:45 "Program Files" isn't in the FAT spec either May 29 22:21:59 ... ? May 29 22:22:16 that's an OS spec May 29 22:22:20 so they could have made a combination of choices resulting in a readable short name May 29 22:22:31 even if they wanted a pretty mixed-case "long" name May 29 22:22:36 I mean, most stuff inside Program Files will have long names anyway May 29 22:22:49 not in 1995 May 29 22:23:00 almost everything still had 8.3 names May 29 22:23:06 really? lol May 29 22:23:36 why would the software developers rename all the files overnight when win95 was released? May 29 22:25:15 I guess if you encountered PROGRA~1 a lot after the introduction of long filename support then that's the real thing to complain about, not the name itself :P I can understand why they didn't limit themselves to a short name, and I actually think "Program Files" is more appropriate than "Programs" May 29 22:25:37 but I imagine this might be a pet peeve if you've run into this a lot in those days May 29 22:25:38 but it does contain programs May 29 22:26:07 in the Swedish version of win95, it's called "Program" May 29 22:26:10 ... and their associated files. they're not where the user would typically go for programs, which is the start menu May 29 22:26:14 singular and plural are the same May 29 22:26:21 *shrug* May 29 22:26:48 well, I ran into PROGRA~1 enough that it was bloody annoying May 29 22:26:54 especially when you had to type it May 29 22:27:49 still not as annoing as macs at the time May 29 22:27:53 like, I think it shouldn't have mattered, they probably thought the same and didn't see good reason to stick with 8.3. I also understand that apparently reality didn't quite work out that way, and annoyance resulted May 29 22:28:08 I don't remember going near a macos9 machine once without it crashing May 29 22:28:28 but that only makes it a bad decision with hindsight, which is not enough to declare it idiotic May 29 22:28:55 is PROGRA~1 not uglier than PROGRAMS? May 29 22:28:56 I don't remember frequent crashes.. what were you doing with them? :P May 29 22:29:03 looking at them, mostly May 29 22:29:09 never really had time for more May 29 22:29:13 they crashed May 29 22:29:19 PROGRA~1 should not have ended up in the user's face, so it doesn't matter how it looks May 29 22:29:27 but it did May 29 22:29:33 a lot May 29 22:29:44 that's the problem, not the fact that it was PROGRA~1 May 29 22:29:47 :P May 29 22:30:04 ah right May 29 22:30:13 because you still had old tools you wanted to use May 29 22:30:19 on stuff in that directory May 29 22:30:23 hmm May 29 22:30:30 some idiotic applications would even open the stock file choooser which would show the long names May 29 22:30:48 then when you selected it, the app would "validate" the name and scream at you for using more than 8 letters May 29 22:31:05 only workaround was to manually type progra~1 May 29 22:31:27 insted of selecting "program files" with the mouse May 29 22:31:34 how's that for friendly? May 29 22:31:55 I don't think there was a good solution for this mess without some transition pain May 29 22:32:17 they could have avoided long names for standard system paths May 29 22:33:42 as for (pre-osx) macs crashing, it is true that lack of memory protection made it more likely that a software crash would take down the whole machine. at the same time, it made the mac a great platform for music software and real-time applications since (if you knew what you were doing) you could literally hook into hardware timers and such May 29 22:36:42 but as a long-time user of macs in the pre-osx time, my experience was that they were great systems... I don't remember annoyance with frequent crashes or such May 29 22:37:07 maybe you ran mostly good software May 29 22:37:11 (after the switch to osx I started using it more like a unix system with a glossy UI, and I quickly realized it was a crappy unix system and I switched to linux) May 29 22:37:52 I mean, that makes sense... why would you use bad software ;) May 29 22:38:00 didn't some of those things come from cpm? May 29 22:38:10 which things? May 29 22:38:25 filenames May 29 22:38:26 sometimes you have to use whatever is on the provided computer May 29 22:38:34 or restrictions thereof May 29 22:38:37 ds2: 8.3 probably did May 29 22:38:48 and the \ May 29 22:38:54 CP/M used 8.3 yeah May 29 22:39:57 https://en.wikipedia.org/wiki/CP/M#Legacy May 29 22:53:02 It May 29 22:53:18 is amazing how long CP/M lingered into today. May 29 22:53:36 windows still has the fractured file system. May 29 22:55:13 It just shows how foolish the SW idea that you can change anything when you want really is. May 29 22:56:47 I mean, it is not false, but it depends on wanting the change more than wanting compatibility :P May 29 23:04:29 windows does really use FAT32 May 29 23:04:40 doesnot May 29 23:05:16 yeah. VFAT May 29 23:05:25 or something? May 29 23:05:28 NTFS May 29 23:05:32 oh May 29 23:05:33 I mean May 29 23:05:38 it can use either May 29 23:05:48 obviously win95 didn't use ntfs yet :P May 29 23:05:51 you can still install win10 with fat? May 29 23:05:57 win95 is not NT based May 29 23:06:29 why are we suddenly talking about filesystems on modern windows systems? May 29 23:06:34 :D May 29 23:06:44 I'll just flowing with it while i work on radar :D May 29 23:13:39 interesting apparently FAT on Windows NT supported both all-uppercase and all-lowercase filename and extension (independently) using two bits in a reserved byte, though neither linux nor Win95 supports this so they'll interpreter it as uppercase May 29 23:14:05 I wonder whether current Windows versions still understand this extension, being NT descendants May 29 23:14:27 oh wait maybe linux does May 29 23:14:47 it totally does May 29 23:32:33 ofc it does :p May 29 23:32:35 :D May 29 23:35:44 and yeah current windows apparently accepts the NT extension. I wonder if it still produces it though or if they disabled that to be backwards compatible with Win95/98/ME May 29 23:38:03 do people really have things like makefile and Makefile in the same directory and expect them to be unique? May 29 23:38:31 I HATE PYTHON SO VERY VERY VERY MUCH May 29 23:40:24 yeah I created py-uio as python library with the idea of making it more accessible to new users... but the more I used python and learned about it, the less I like it May 29 23:43:44 yeah libwww-perl used to include http commandline utilities called HEAD, GET, and POST May 29 23:45:10 which resulted in some people managing to overwrite /usr/bin/head on MacOS X .. they should have installed into /usr/local anyway :P May 29 23:59:54 apt install qt5-default still works on Debian Buster. Heads up! May 30 00:00:09 ... why would it not? May 30 00:00:31 I have been looking to install a backend for a bit now. I just came across this item in the google groups. May 30 00:01:01 Why would it not? Well, maybe b/c I have been trying w/ other ways outside of apt. May 30 00:01:39 apt should be the obvious first thing to try to get software installed May 30 00:01:54 especially for something as big and complicated as qt May 30 00:02:15 Right...but sometimes the versioning is outdated. And yes, you are right about qt. I should have picked apt firstly. May 30 00:02:44 debian tends to favor stability over having the latest versions of everything yeah May 30 00:03:14 I noticed. It is smart but some of the libraries I want to use are calling for specific versions of different libraries. May 30 00:05:21 I need pyqt5 or something similar. May 30 00:05:35 I have not been able to find out how to install them just yet. May 30 00:05:52 Let me try apt. May 30 00:06:20 apt install python3-pyqt5 May 30 00:06:30 Dang! May 30 00:06:33 You are right. May 30 00:06:47 This is exactly what I have been trying to install. May 30 00:06:49 Nice. May 30 00:07:13 I also tried wxPython but that failed w/ the "recommended" way. May 30 00:09:37 python3-wxgtk4.0 May 30 00:10:59 Nice. May 30 00:11:42 I will try that. My error reads: Cannot import QtOpenGL from PyQt5. May 30 00:11:48 Dang. May 30 00:12:30 apt install python3-pyqt5.qtopengl May 30 00:12:49 Dang. @zmatt: Where are you searching for these items? May 30 00:13:23 I look on the debian site but there search engine usually gives me troubles. May 30 00:13:50 apt-cache search pyqt5 May 30 00:13:56 Oh! May 30 00:14:07 apt-cache search. Thank you. Yikes. May 30 00:14:17 set_ possible video answers https://pastebin.com/HY5rEzwF May 30 00:15:55 Nice. Thank you. I am about to run a simple frame catcher from webcam script. May 30 00:16:56 aborted. Dang. May 30 00:17:16 The system seemed to work and then it aborted my source. May 30 00:17:42 I am using imageio and visvis. May 30 00:18:05 Wait for the odd and peculiar error. May 30 00:19:38 https://pastebin.com/8ANznHh2 is the error. Thank you fellows. GenTooMan: I saw the paste. May 30 00:20:33 Where do you think I should start looking for this type of error? May 30 00:23:17 oh wait, you're trying to use OpenGL on a beaglebone? May 30 00:24:24 Sort of. May 30 00:24:47 Let me go and make sure again. May 30 00:24:54 maybe mesa's software renderer doesn't support GL_POINT_SMOOTH May 30 00:26:23 pyopengl. May 30 00:26:30 Oh. May 30 00:26:31 Okay. May 30 00:27:34 I will go and look. May 30 00:27:55 probably a lot of OpenGL programs just wouldn't bother checking the return code on that, hence would end up simply not using the GL_POINT_SMOOTH feature... but this python3 library turns all errors into exceptions, thus terminatin the program is the exception isn't handled by the programmer May 30 00:28:21 Right but the webcam still runs. May 30 00:28:33 The program stops and the webcam stays on. May 30 00:28:45 Some odd stuff. Google got involved a while back. May 30 00:28:53 I mean, this has nothing to with a cam May 30 00:28:59 Okay. May 30 00:29:41 So, I guess I can go in to the source and stop the GL_POINT_SMOOTH feature. May 30 00:29:45 That might work. May 30 00:32:21 It might be legacy too. May 30 00:34:15 Well, it is a boolean or GLboolean. May 30 00:34:33 So, I guess I could call False instead of True. May 30 00:36:53 glDisable (GL_POINT_SMOOTH); <<< Maybe? May 30 00:37:13 no, omit the call entirely May 30 00:37:31 Oh. Okay. Off too look for the source on my system. May 30 00:37:41 but honestly, opengl software rendering on a beaglebone is going to _awful_ May 30 00:37:55 That is fine. I am just catching a frame of video. May 30 00:37:56 you should really reconsider the path you're on May 30 00:38:12 I am just learning for learning's sake. May 30 00:38:20 you're using a visualiation library (visvis) that uses opengl May 30 00:38:20 Nothing more. May 30 00:38:25 Right. May 30 00:38:36 that's going to be slow as fuck May 30 00:39:06 Oh. OKay. Is there another option? May 30 00:40:42 I have no idea what you're even trying to do, nor do I really want to think about it. you seem to be trying to lot a whole bunch of things that the BBB is really bad at May 30 00:40:54 Oh. Okay. May 30 00:40:57 No issue. May 30 02:07:32 lol, as a way to convey how little energy 1 mbmbam (millibar-millibarn-attometer, i.e. 10^-47 J) is xkcd used the example "roughly the energy you'd need to lift one yeast cell by one planck length in Earth's gravity." while someone over at the explainxkcd wiki came up with "the work necessary to pull two socially distancing (6 ft) SARS-CoV-2 virions apart by the thickness of a single strand of hair ... May 30 02:07:38 ...against the gravity they exert on each other." May 30 02:16:58 Evening May 30 02:18:05 What is going on with these riots and looters and burning and now attacking police.. May 30 02:23:22 I stopped watching/reading news quite some time ago, I feel like it's made me a happier person May 30 02:25:57 Party? May 30 02:28:54 set_ What is going on with these riots and looters and burning and now attacking police.. May 30 02:29:48 I am afraid the National Guard my start shooting... May 30 02:43:48 I do not know. May 30 02:44:00 Probably about the guard. May 30 02:44:17 Now, if there is a way to segway a BBB in it, collective. May 30 02:44:38 "set_ started recording the riots w/ his BBB!" May 30 02:45:10 I am close on my OpenGL. It is not OpenGL issues any longer. I am on visvis only now. May 30 02:45:21 Well, w/ errors. **** ENDING LOGGING AT Sat May 30 02:59:59 2020