**** BEGIN LOGGING AT Wed Jun 24 02:59:57 2020
Jun 24 03:35:46 <shwaank> Generic question here. I submitted a pull request that got merged some time ago, but still doesn't get ingested on any of the releases package feeds other than 'snapshot'. Do I need to submit another pull request to up the versioning in the makefile in order for it to get taken?
Jun 24 03:37:31 <lmore377> shwaank: If I remember correctly, it will me merged into release when the next one comes out so you'll just need to play the waiting game
Jun 24 03:38:57 <shwaank> its been a few releases already? https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=9105057cc0da6172a77490733769a0e5f3c7b2d2
Jun 24 03:40:17 <lmore377> Oh I'm not sure then, I just remember reading that somewhere
Jun 24 03:43:32 <owrt-snap-builds> build #423 of kirkwood/generic is complete: Failure [failed defconfig dltar]  Build details are at http://buildbot.openwrt.org/master/images/builders/kirkwood%2Fgeneric/builds/423  blamelist: Catalin Patulea <catalinp@google.com>, Leon M. George <leon@georgemail.eu>, Alex Lewontin <alex.c.lewontin@gmail.com>, Sungbo Eo <mans0n@gorani.run>, Edward
Jun 24 03:43:32 <owrt-snap-builds> Matijevic <motolav@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, Sven Roederer <devel-sven@geroedel.de>, Jose Olivera <oliverajeo@gmail.com>
Jun 24 04:48:27 <owrt-snap-builds> build #377 of bcm53xx/generic is complete: Failure [failed kmodconfig]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm53xx%2Fgeneric/builds/377  blamelist: Catalin Patulea <catalinp@google.com>, Leon M. George <leon@georgemail.eu>, Alex Lewontin <alex.c.lewontin@gmail.com>, Sungbo Eo <mans0n@gorani.run>, Edward Matijevic
Jun 24 04:48:27 <owrt-snap-builds> <motolav@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, Sven Roederer <devel-sven@geroedel.de>, Jose Olivera <oliverajeo@gmail.com>
Jun 24 04:50:01 <owrt-snap-builds> build #364 of malta/be is complete: Failure [failed defconfig dltar]  Build details are at http://buildbot.openwrt.org/master/images/builders/malta%2Fbe/builds/364  blamelist: Catalin Patulea <catalinp@google.com>, Leon M. George <leon@georgemail.eu>, Alex Lewontin <alex.c.lewontin@gmail.com>, Sungbo Eo <mans0n@gorani.run>, Edward Matijevic
Jun 24 04:50:01 <owrt-snap-builds> <motolav@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, Sven Roederer <devel-sven@geroedel.de>, Jose Olivera <oliverajeo@gmail.com>
Jun 24 04:52:00 <owrt-snap-builds> build #358 of ath79/generic is complete: Failure [failed defconfig dltar]  Build details are at http://buildbot.openwrt.org/master/images/builders/ath79%2Fgeneric/builds/358  blamelist: Catalin Patulea <catalinp@google.com>, Leon M. George <leon@georgemail.eu>, Alex Lewontin <alex.c.lewontin@gmail.com>, Sungbo Eo <mans0n@gorani.run>, Edward Matijevic
Jun 24 04:52:00 <owrt-snap-builds> <motolav@gmail.com>, Adrian Schmutzler <freifunk@adrianschmutzler.de>, Sven Roederer <devel-sven@geroedel.de>, Jose Olivera <oliverajeo@gmail.com>
Jun 24 05:38:57 <lmore377> is there anyone here that can review pull requests on github? adrianschmutzler was reviewing my request and it's been 4 days since his last reply. the pull request is https://github.com/openwrt/openwrt/pull/3115
Jun 24 05:43:45 <lmore377> also i want to add my device to the toh but the manufacturer (Askey) isn't listed
Jun 24 06:04:58 <pkgadd> lmore377: for the wiki (and adding a new manufacturer), you need to ping a moderator (e.g. tmomas) in the forum
Jun 24 06:06:16 <lmore377> pkgadd: oh alright thanks
Jun 24 07:27:35 <lmore377> does openwrt automatically shrink the size of the last mtd partition if it detects that it wont phisically fit in nand?
Jun 24 07:50:53 <blogic> we use dynamic volumes inside ubi
Jun 24 07:51:11 <blogic> so the volumes will be sized to whatever is needed and grow as more space is required
Jun 24 08:05:42 <lmore377> so ive been porting to a new router and i have 2 of them and they're identical except one has a smaller flash. ive been using the one with bigger flash so if i flash the image for that one to the one with the smaller flash, will it be fine?
Jun 24 08:21:20 <blogic> how do you flash ?
Jun 24 08:21:58 <blogic> i use sysupgrade-tar on nand devices and that will make owrt use dynamic volumes
Jun 24 08:22:34 <lmore377> i do it by booting an initramfs image and using sysupgrade
Jun 24 08:22:48 <blogic> yes but what image type do you flash ?
Jun 24 08:22:54 <blogic> ubi ubifs tar
Jun 24 08:24:04 <lmore377> oh tar but after flash it uses ubi automatically
Jun 24 08:24:10 <blogic> yes
Jun 24 08:24:13 <blogic> then it will work
Jun 24 08:25:27 <lmore377> oh nice. i was planning on only supporting the one with bigger flash but looks like i'll just do both. i'll do a bit of testing forst anyways
Jun 24 08:25:38 <lmore377> first*
Jun 24 08:37:55 <rsalvaterra> I just noticed… has the overlay filesystem changed from 19.07 to master? I'm looking at the mount output now, and it's ext4. I was almost sure it was f2fs before.
Jun 24 08:40:33 <rsalvaterra> Or maybe I just need coffee.
Jun 24 08:41:00 <rsalvaterra> (This is on both mvebu and x86-64.)
Jun 24 08:44:09 <jow> rsalvaterra: iirc f2fs vs. ext4 depends on the available size
Jun 24 08:44:33 <jow> there's a switch somewhere that selects f2fs for small and ext4 for big overlays
Jun 24 08:45:11 <rsalvaterra> Hmm… is 100 MB "big"?
Jun 24 08:45:23 <rsalvaterra> From my Omnia: /dev/loop0               91.2M      1.6M     82.8M   2% /overlay
Jun 24 08:46:25 <jow> rsalvaterra: https://lxr.openwrt.org/source/fstools/libfstools/rootdisk.c#L110
Jun 24 08:46:30 <jow> rsalvaterra: #define F2FS_MINSIZE            (100ULL * 1024ULL * 1024ULL)
Jun 24 08:46:37 <jow> so exactly 100MB as it seems :)
Jun 24 08:47:24 <rsalvaterra> *facepalm*
Jun 24 08:48:33 <rsalvaterra> I'd rather force it to be f2fs always, even for "smart" eMMC block devices. Their wear leveling is usually not that smart.
Jun 24 08:51:21 <jow> I forgot about the original reasoning
Jun 24 08:51:58 <rsalvaterra> I'll dig into the git history. ;)
Jun 24 08:52:07 <jow> iirc it was either that f2fs was slow for big blockdevs or that it performed badly [compared to ext4] on traditional blockdevs (spinning disk sata / ssd)
Jun 24 08:53:55 <rsalvaterra> If you need to do so many reads/writes on the overlay to notice that, your main problem is not the filesystem performance. :P
Jun 24 08:54:32 <jow> well on x86 I
Jun 24 08:54:49 <jow> 'd prefer some plain boring standard ex4 over f2fs
Jun 24 08:55:34 <rsalvaterra> True, I forgot x86 with large block devices was a thing.
Jun 24 08:56:45 <rsalvaterra> But ext4 with journaling isn't really the healthiest on flash.
Jun 24 08:57:25 <jow> that's likely the reason for the <100MB heuristic
Jun 24 08:58:08 <jow> not sure if there's other way to figure out if a blockdev is emmc
Jun 24 08:58:21 <jow> or some other flash-like medium
Jun 24 08:58:54 <rsalvaterra> I belive there's a "non-rotational" property somewhere in sysfs.
Jun 24 08:59:00 <rsalvaterra> *believe
Jun 24 09:00:26 <rsalvaterra> Yep! It's called "rotational".
Jun 24 09:00:52 <rsalvaterra> root@heimdal:~# cat /sys/block/mmcblk0/queue/rotational 0root@heimdal:~#
Jun 24 09:01:18 <rsalvaterra> Gah! I need a new IRC client.
Jun 24 09:05:10 <rsalvaterra> But this is MMC, I don't know if it's the same on MTD.
Jun 24 09:13:39 <rsalvaterra> Oh, wait, F2FS_MINSIZE! No biggie for me, I'll just increase the overlay size. I guess what happened was that the default overlay size was reduced on the defconfig.
Jun 24 09:14:15 <jow> its confusingly named
Jun 24 09:14:40 <jow> it should be F2FS_MAXSIZE (when smaller than that, use f2fs) or EXT4_MINSIZE (when bigger than that, use ext4)
Jun 24 09:15:50 <rsalvaterra> Not if I'm reading the code correctly…
Jun 24 09:16:05 <rsalvaterra> 17         if (ioctl(fd, BLKGETSIZE64, &size) == 0)118                 ret = size - p->offset > F2FS_MINSIZE;
Jun 24 09:16:30 <rsalvaterra> I uses f2fs if the size is greater than F2FS_MINSIZE.
Jun 24 09:16:30 <jow> ah indeed
Jun 24 09:16:34 <jow> odd
Jun 24 09:16:39 <jow> *shrug*
Jun 24 09:17:24 <rsalvaterra> Come to think of it, it's not that odd… f2fs has a hefty amount of metadata overhead.
Jun 24 09:18:59 <jow> 12:13 < nbd> i think i will add ext4 as an option too, because f2fs cannot deal with small partitions easily
Jun 24 09:19:02 <jow> 12:21 < nbd> ext4 is more common, and it's only going to be used for the corner case if the overlay partition is smaller than something like 100 MB
Jun 24 09:19:05 <jow> 12:21 < nbd> everything bigger will use f2fs
Jun 24 09:19:05 <jow> from my irc log archive
Jun 24 09:19:34 <jow> so yeah, seems to be as intended and you simply need to increase your overlay
Jun 24 09:19:43 <rsalvaterra> Indeed! I noticed this on my Eee PC 901 (which is running Debian Sid from an SD card; crazy, I know). I allocated 128 MiB to the boot partition and the metadata overhead is almost 40 MiB.
Jun 24 09:20:28 <jow> I somehow thought that f2fs is something like jffs2, some weird niche fs for small flashes
Jun 24 09:20:42 <jow> but it's rather a flash friendly general purpose fs
Jun 24 09:21:16 <rsalvaterra> Exactly, it's general purpose.
Jun 24 09:21:36 <rsalvaterra> I can format spinning rust with it. Not a great idea, though.
Jun 24 09:28:34 <rsalvaterra> Ooh, I see zx2c4 already merged the compat WireGuard update! I'm going to build an image without the kernel Thumb-2 relocation workaround, to test it. (This doesn't invalidate my previous patch, I can submit a follow-up later on.)
Jun 24 09:41:58 * ldir has the remaining 15GBish of the 16GB SSD card thingy formatted as f2fs on his apu2
Jun 24 09:46:00 <stintel> I think I'm using ext4
Jun 24 09:46:16 <stintel> mounted it on /srv and use it for my LXC pi-hole container
Jun 24 09:51:15 <rsalvaterra> I never really understood the point of pi-hole.
Jun 24 09:51:58 <jow> dns blocking with statistics
Jun 24 09:52:06 <jow> and simple configuration
Jun 24 09:52:20 <rsalvaterra> Yeah, but… adblock on OpenWrt…?
Jun 24 09:52:37 <rsalvaterra> Sure, no statistics and not pretty GUI.
Jun 24 09:52:41 <rsalvaterra> *no
Jun 24 09:52:45 <jow> for many it is simpler to use a dedicated raspi appliance preconfigured for adblocking than to install openwrt in the first place
Jun 24 09:52:53 <jow> and redo the entire home network while being at it
Jun 24 09:52:57 <jow> and find a supported router
Jun 24 09:54:40 <stintel> I would be able to do the same with adblock? interesting. I kind of dislike the fact that pi-hole is installer based
Jun 24 09:54:46 <rsalvaterra> I'm using stubby with AdGuard DNS. Pretty happy with it! :)
Jun 24 09:54:49 <stintel> main reason for running it in a container
Jun 24 09:55:21 <stintel> not to mention an sqlite db in bloody /etc
Jun 24 09:55:43 <rsalvaterra> stintel: dear God, what are you doing with your life? :D
Jun 24 09:55:49 <jow> I think the only "interesting" bit in pi-hole is the patched dnsmasq to expose query stats
Jun 24 09:56:11 <stintel> rsalvaterra: ?
Jun 24 09:56:14 <jow> plus a fancy ui, but that is simple to build once the bits and pieces are there
Jun 24 09:56:34 <rsalvaterra> Yeah, I agree the pretty stats are useful, but to me it's just overhead… also the reason I don't use LuCI.
Jun 24 09:57:36 <rsalvaterra> And I only stopped using adblock because I could get the same or similar results with AdGuard DNS, without the local overhead.
Jun 24 09:58:14 <rsalvaterra> Of course that means trusting your upstream servers. Not everyone will be happy with that.
Jun 24 10:01:17 <rsalvaterra> jow: I think adblock can also provide stats with tcpdump.
Jun 24 10:01:34 <jow> the overhead of tcpdump is massive
Jun 24 10:02:08 <rsalvaterra> https://git.openwrt.org/?p=feed/packages.git;a=blob;f=net/adblock/files/README.md;h=da834bd4a3ba82d1257e7176a01c8ee9536646c0;hb=HEAD#l104
Jun 24 10:03:14 <rsalvaterra> Yeah, I'm not arguing that.
Jun 24 10:04:38 <rsalvaterra> I can't live without tcpdump(-mini). It's the only way I can get WireGuard working correctly from scratch.
Jun 24 10:07:59 <karlp> that's not exactly a ringing endorsement of wireguard :)
Jun 24 10:11:53 <rsalvaterra> karlp: Nah, it's just me. Sometimes I screw up the configuration. WireGuard itself is awesome.
Jun 24 10:15:18 <rsalvaterra> My only issue with it is related to fascist sysadmins who block UDP on their networks. Especially public Wi-Fi networks.
Jun 24 10:16:49 <karlp> if you need tcpdump to figure out a config problem, I'd say the built in diags are.... lacking?
Jun 24 10:18:59 <rsalvaterra> We're heading into philosophical territory… ;)
Jun 24 10:20:21 <rsalvaterra> I believe WireGuard is engineered to be as simple as possible.
Jun 24 10:21:11 <rsalvaterra> And if you can debug any problems with tcpdump, why bloat it with redundant internal diagnostics?
Jun 24 10:28:57 <rsalvaterra> And regarding the UDP issue I mentioned, it's not impossible to work around it… https://github.com/wangyu-/udp2raw-tunnel
Jun 24 10:30:56 <stintel> I can't live without tcpdump on any networking device period
Jun 24 10:31:53 <stintel> tracked down an SSL bug in the Jetbrains Upsource Docker container yesterday with the help of it
Jun 24 10:32:19 <rsalvaterra> Yes! It's *that* useful. I may run out of space for other stuff on my 8 MiB flash routers, but tcpdump stays.
Jun 24 10:33:13 <stintel> this is related: https://webtide.com/jetty-alpn-java-8u252/ so the JDK was updated to 8u252 in the container but they kept the Jetty ALPN libraries (1.8). conflicting with the now backported ALPN APIs
Jun 24 10:33:29 <stintel> fixed it by mounting an empty volume on the directory containing the Jetti ALPN libs :P
Jun 24 10:34:13 <stintel> because the wrapper and launcher scripts is a horrible mess in which I could not disable the loading of those libs
Jun 24 10:34:22 <stintel> Java project FTL
Jun 24 10:34:26 <stintel> projects*
Jun 24 10:36:56 <rsalvaterra> Jeez, don't get me started. I'm a Java developer for a living and I feel like I'm getting dumber because of it.
Jun 24 10:37:21 <jow> just add more jars
Jun 24 10:37:35 * rsalvaterra screams
Jun 24 10:37:46 <jow> :)
Jun 24 10:38:05 <stintel> :P
Jun 24 10:38:37 <rsalvaterra> Java developer working for the insurance industry. I mean, how low can you go?
Jun 24 10:39:23 <stintel> hey if it pays the bills ...
Jun 24 10:39:36 <rsalvaterra> True, stintel…
Jun 24 10:39:58 <rsalvaterra> But dealing with all that legacy…
Jun 24 10:40:08 <stintel> we're developing mostly Java software also
Jun 24 10:40:20 <stintel> but I'm in an ops position, not a developer really
Jun 24 10:41:57 <jow> I always disliked Java from an ops perspective because it was so... opaque
Jun 24 10:42:03 <rsalvaterra> Once I ended up trying to figure out a bug on a COBOL program (yeah, we invoke COBOL routines from Java)…
Jun 24 10:42:13 <rsalvaterra> … which was written before I was born.
Jun 24 10:42:31 <jow> you had to load weird .jar's into it to extract runtime info... or talks some arcane jmx console command protocol
Jun 24 10:44:06 <karlp> discovering jmx was amaze, like, where has this been all my life.
Jun 24 10:44:20 <karlp> got an instant ui with jbench stuff for tweaking knobs in my apps
Jun 24 10:44:33 <karlp> don't realyl miss doing java for a day job though...
Jun 24 10:44:40 <rsalvaterra> The thing is, I'm 39. That code was written in the 70's.
Jun 24 10:45:45 <stintel> I decided to enroll in Harvard's CS50x. first task I need to do is build an app in Scratch. Blockly like GUI programming. oh dear what did I get myself into. I think I'd prefer doing it in Java even :P
Jun 24 10:46:42 <stintel> mainly doing it be cause I keep hitting walls when I'm trying to do OpenWrt stuff and it really started to frustrate me
Jun 24 10:46:47 <rsalvaterra> Don't knock Scratch, it's *fantastic* to get kids into programming!
Jun 24 10:47:09 <stintel> rsalvaterra: agreed. but I am 35 and have written some code. GUI is just ugh for me
Jun 24 10:47:48 <rsalvaterra> I wish I had Scratch the first time I fired up GWBASIC… ;)
Jun 24 10:49:23 <stintel> Borromini: namesake  o/
Jun 24 10:49:35 <Borromini> stintel: sup
Jun 24 10:49:43 <stintel> SSDD :P
Jun 24 10:49:58 <rsalvaterra> Woohoo! Success. The WireGuard module loaded without the kernel Thumb-2 workaround! \o/
Jun 24 10:50:28 <stintel> yay
Jun 24 10:51:07 <stintel> I've tried wireguard once when I was staying in a hotel end of 2017. didn't manage to get it up and running fast enough to my liking. so since then I'm still relying on my IPsec setups
Jun 24 10:51:22 <stintel> maybe one day :)
Jun 24 10:52:03 <rsalvaterra> IPSec…? Before WireGuard as a thing, I started reading on how to configure it on OpenWrt and ran away screaming.
Jun 24 10:52:15 * ldir lols
Jun 24 10:53:03 <rsalvaterra> Even OpenVPN is a nuissance.
Jun 24 10:53:05 <stintel> I'm maintaining strongswan in the package feeds. documented it for my linux using colleagues. both old ipsec.conf and new swanctl ways. also made a mobileconfig for all the apple users and wrote powershell script to set it up on Windows
Jun 24 10:53:23 <stintel> and it's still a mystery to me :P
Jun 24 10:54:39 <stintel> but it works and I like that you can configure it in a way that iOS, OS X, Windows 7+ all can connect to it with the native VPN client
Jun 24 10:55:31 <rsalvaterra> I'm sure it's just a matter of time before WireGuard goes native on all operating systems, but I understand the point. ;)
Jun 24 10:55:32 <Borromini> stintel: hehe :P
Jun 24 10:55:41 <Borromini> wireguard is neat.
Jun 24 10:56:23 <Borromini> i have a tunnel set up for my brother's home office on windows (openwrt 'server') and here at home it's all linuxes (and openwrt 'server')
Jun 24 10:56:32 <rsalvaterra> WireGuard "saved my life" when I travelled to Brazil. From my phone's point of view, I never left home. :P
Jun 24 10:56:42 <Borromini> lol
Jun 24 10:57:01 <stintel> in what way? hiding from your employer that you were in a different country? :P
Jun 24 10:57:23 <jow> ipsec? https://www.schneier.com/academic/paperfiles/paper-ipsec.pdf
Jun 24 10:58:06 <jow> money quote: "we do not believe that it will ever result in a secure operationalsystem. It is far too complex, and the complexity has lead to a large numberof  ambiguities,  contradictions,  inefficiencies,  and  weaknesses.  It  has  been  veryhard work to perform any kind of security analysis; we do not feel that we fullyunderstand the system, let alone have fully analyzed it."
Jun 24 10:58:09 <rsalvaterra> Nah, I'm just lazy and couldn't be bothered to to put up with all the extra checks when I had to log on to the banking apps.
Jun 24 10:58:12 <jow> pretty much sums it up :P
Jun 24 10:58:56 <stintel> ah I usually travel back and for between Bulgaria and Belgium but my different banks in different countries have no checks for this
Jun 24 10:59:03 <stintel> maybe because I stay inside of EU though
Jun 24 10:59:07 <Borromini> jow: lol.
Jun 24 10:59:08 <stintel> jow: ;)
Jun 24 11:00:09 <rsalvaterra> Possibly. In my case, I travelled from Portugal, so I guess it triggered all the alarms. :P
Jun 24 11:01:25 <stintel> ah well I've set up my IPsec tunnels way before wireguard came to be. roadwarrior configs on all my locations (.be .bg .fr - OVH)
Jun 24 11:01:45 <stintel> and S2S between those locations too
Jun 24 11:02:10 <rsalvaterra> Before WireGuard, I used an SSH tunnel… ;)
Jun 24 11:02:14 <stintel> :P
Jun 24 11:02:30 <stintel> used to do OpenVPN but that feels like a lifetime ago
Jun 24 11:05:59 <rsalvaterra> jow: my overlay is still ext4, I guess sysupgrade won't cut it…? I increased the size of the partition to 512 MiB, but it wasn't reformatted as f2fs.
Jun 24 11:07:24 <ldir> I started with a bit of Python recently https://www.py4e.com/lessons - Being 'self taught' in the early 80's home computer boom and then the college course I took was absolutely shit I always feel like I 'missed out' on some fundamental CS things and also hit walls as a result
Jun 24 11:07:58 <Borromini> ldir: prior knowledge required or absolute basics?
Jun 24 11:08:06 <Borromini> i still have an o'reilly book on python.
Jun 24 11:08:17 <Borromini> ok 'no programming background' :P
Jun 24 11:09:12 <stintel> ldir: python is my goto language for anything that is too ugly todo in bash :P
Jun 24 11:09:24 <rsalvaterra> I never really tried Python… I learned Lua in an evening, though. :P
Jun 24 11:09:37 <ldir> I look at the stuff that I used (sinclair zx80/zx81) in amazement at how clever the programmers were and what they did in 4K/8K of rom.  I've read stuff about the Apollo guidance computer in wonderment as well.
Jun 24 11:10:17 <rsalvaterra> (I was bored and decided to do some fancy stuff with the leds on my Omnia.)
Jun 24 11:10:18 <karlp> I sometimes wonder if they made it work, because they only had that to work with
Jun 24 11:10:32 <karlp> instead of 15 gajilliion options, all loosely compatible and incomaptible
Jun 24 11:10:37 <ldir> I feel like (but may be wrong) that the golden age/really fundamental clever stuff was done in the 60s
Jun 24 11:10:52 <rsalvaterra> ldir: have you watched the AGC restoration project, on YouTube?
Jun 24 11:11:50 <ldir> not sure, I watched a CCC ultimate agc talk and read Don Eyles book 'sunburst and luminary'
Jun 24 11:12:03 <rsalvaterra> I also watched the CCC talk, but…
Jun 24 11:12:12 <rsalvaterra> … go watch this right now. ;)  https://www.youtube.com/watch?v=2KSahAoOLdU&list=PL-_93BVApb59FWrLZfdlisi_x7-Ut_-w7
Jun 24 11:12:33 <ldir> I'm going to feel really stupid aren't I :-)
Jun 24 11:13:16 <rsalvaterra> Not at all, you already have the full background!
Jun 24 11:15:16 <ldir> Borromini: I found the python for everyone course interesting and fun.  I haven't finished it yet.  I had a mind fu*k at 'tuples' but that turned out to be a language barrier/misunderstanding/otherthinking thing in my brain
Jun 24 11:18:25 <rsalvaterra> I had a similar experience with Lua, when I learned that a function could return multiple values. :P
Jun 24 11:18:40 <rsalvaterra> "WAT"
Jun 24 11:28:45 <rsalvaterra> ldir: you're right about the really clever stuff having been developed in the 60's.
Jun 24 11:29:52 <rsalvaterra> Arguably, there's no arquitectural improvement in CPU's since that time. Only process improvements.
Jun 24 11:32:24 <rsalvaterra> In other words, you have a miniature System/360 model 91 in your pocket. ;)
Jun 24 11:33:54 <rsalvaterra> Caches, floating-point, superscalar pipelining, OoO execution, multi-channel memory… it's all there.
Jun 24 11:38:19 <dwmw2_gone>  rsalvaterra nah, CPUs came up with all this speculative stuff since then...
Jun 24 11:38:21 <dwmw2_gone> hm, ok.
Jun 24 12:41:19 <rsalvaterra> jow: Crap! I was bitten by the sysupgrade. That's why my overlay was stil formatted as ext4.
Jun 24 12:42:25 <rsalvaterra> I don't know why, for some reason, sysupgrade doesn't work on my Omnia. The system is unchanged, I have to repeat the process until it eventually upgrades (usually once, rarely twice).
Jun 24 12:43:04 <rsalvaterra> I'd say the first sysupgrade works about 80 % of the time.
Jun 24 12:44:37 <rsalvaterra> So I actually hadn't really tested the Thumb-2 fix… but now I did, and it's working 100 %, as expected.
Jun 24 12:44:42 <rsalvaterra> [    8.691783] wireguard: WireGuard 1.0.20200623 loaded. See www.wireguard.com for information.
Jun 24 15:28:49 <owrt-snap-builds> build #426 of mediatek/mt7623 is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/mediatek%2Fmt7623/builds/426
Jun 24 16:22:59 <karlp> man, have a uloop+inotify+lua thing.  doesn't get any followup inotify events.
Jun 24 16:23:27 <karlp> run it against current libubox+lua uloop, works fine.
Jun 24 16:23:49 <karlp> check files on disk, jan 21, checkout there, still runs fine.
Jun 24 16:24:02 <karlp> and no changes of relevance to lua or uloop this year anyway
Jun 24 16:25:07 <karlp> and doing things like stripping out uncalled functions is making it work/stop work.
Jun 24 16:25:15 <karlp> smells like memory corruption but yay
Jun 24 16:26:09 <rsalvaterra> I complained about that already… :D
Jun 24 16:26:24 <rsalvaterra> Stripping is causing problems in the userspace.
Jun 24 16:26:31 <rsalvaterra> (In master.)
Jun 24 16:47:10 <karlp> I asked you to clarify that and you said it was happeningw ithout sstrip anyway
Jun 24 16:47:25 <karlp> and this is also happening with a locally built libubox and local lua on my laptop
Jun 24 16:47:33 <karlp> just not one I built today...
Jun 24 16:47:45 <rsalvaterra> Yeah, I honestly have no idea. Maybe something changed in procd…?
Jun 24 16:50:40 <rsalvaterra> But you've reproduced it too, right?
Jun 24 16:51:04 <karlp> your claimed sstrip problem? no, I couldn't even clarify what you were talking about :)
Jun 24 16:51:43 <karlp> I just have a weird uloop issue that's heisunbugging when I try and look at it.
Jun 24 16:51:57 <karlp> like removing an uncalled lua functino and re-running will change whether inotify works or not...
Jun 24 16:55:04 <adrianschmutzler> archs38 fails to build since Jun 22:
Jun 24 16:55:18 <adrianschmutzler> dbrandom.c:174:36: error: &#39;GRND_NONBLOCK&#39; undeclared (first use in this function); did you mean &#39;SOCK_NONBLOCK&#39;?
Jun 24 16:57:14 <adrianschmutzler> and mpc85xx/generic still needs to be "migrated" to mpc85xx/p1010
Jun 24 17:17:37 <owrt-snap-builds> build #424 of kirkwood/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/kirkwood%2Fgeneric/builds/424
Jun 24 18:07:15 <owrt-snap-builds> build #378 of bcm53xx/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/bcm53xx%2Fgeneric/builds/378
Jun 24 18:39:25 <owrt-snap-builds> build #365 of malta/be is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/malta%2Fbe/builds/365
Jun 24 19:14:24 <jow> rsalvaterra: karlp: mklibs != sstrip
Jun 24 19:14:33 <jow> mklibs is known to cause trouble
Jun 24 19:55:46 <owrt-snap-builds> build #359 of ath79/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/ath79%2Fgeneric/builds/359
Jun 24 21:47:17 <Slimey> https://github.com/Swordfish90/cool-retro-term
Jun 24 22:56:49 <karlp> bleeding-harsh-term :)
Jun 24 22:56:57 * karlp doesn't miss old terminals one bit
Jun 24 23:06:08 <rsalvaterra> Old terminals?
Jun 24 23:08:36 <rsalvaterra> I wouldn't mind hooking up a Teletype Model 33 to my machine, as a console… for science/lulz, of course. :P
Jun 24 23:23:37 <pkgadd> I've ruined my eyes (that's an exaggeration, don't need glasses yet, but I've gotten a zero tolerace for flickering after using CRTs on 60Hz for years) on old CRTs, horrible days
Jun 24 23:52:31 <bemo> running into a build problem with reference code I received from a vendor, and trying to debug it... hoping someone can help me figure out what's happening... long-story-short... I can build the non-ssl version of mosquitto client, but when I try building the ssl version I see errors about not being able to find the libssl library... I decided to
Jun 24 23:52:32 <bemo> try to reproduce this with a stock 18.06.2 tree (with a different architecture), and it build just fine... but when I decided to try to compare the commands used to build... they were similar... and, I discovered if I manually used the command-line to build it in the working openwrt tree, it also failed with the same errors ("ld: warning:
Jun 24 23:52:32 <bemo> libssl.so.1.0.0, needed by ../lib/libmosquitto.so.1, not found (try using -rpath or -rpath-link)")
Jun 25 00:06:50 <owrt-snap-builds> build #457 of arc770/generic is complete: Success [build successful]  Build details are at http://buildbot.openwrt.org/master/images/builders/arc770%2Fgeneric/builds/457
**** ENDING LOGGING AT Thu Jun 25 02:59:57 2020