**** BEGIN LOGGING AT Tue Jun 09 02:59:57 2020 Jun 09 16:21:40 so I am doing a bit of software archeology on opkg, for an Open Source Summit presentation, hoping I could get some more color/corrections from the community Jun 09 16:21:55 so far, I understand that Ipkg was created for the Linksys NSLU2 on 2004, but was not mantained around 2008 Jun 09 16:22:11 so OpenMoko decided to fork it, that's where the 'O' comes from, since there was a trademark on ipkg Jun 09 16:22:33 OpenMoko was using OE, so that's how it ended up as an OE project? Jun 09 16:22:45 adelcast: Hey, I saw your email about that, been meaning to reply. On a call right now but will give some replies soon Jun 09 16:23:55 @hrw any thoughts on thre history of ipkg/opkg? Jun 09 16:24:07 thanks Paul, yeah, I am going to pre-record the session so now I am on a shorter deadline....sight...stupid virus Jun 09 16:24:07 elo Jun 09 16:24:43 adelcast: openmoko system devs were also openembedded devs so nothing strange that opkg ended here Jun 09 16:25:04 adelcast: also there were issues with ipkg Jun 09 16:25:29 as there was ipkg written in shell (mostly used in openwrt) and ipkg written in C (used in OE). Jun 09 16:25:53 ah ok, that was going to be my next question, the openwrt fork Jun 09 16:26:07 at some moment amount of patches to ipkg crossed magical point and it was easier to fork it and get it working better. opkg was born Jun 09 16:26:07 right now, as far as I can tell, they are a fork of OE opkg Jun 09 16:26:30 but my memory can be rusty Jun 09 16:27:08 I think you are right, I saw some old openmoko ml post mentioning that there were just too many patches not being absorbed Jun 09 16:27:35 adelcast: one of changes was also creation of libopkg so it was possible to write GUI for package managers Jun 09 16:27:52 before you had to parse ipkg info and exec() ipkg Jun 09 16:28:20 but it would require digging in source history so probably you know more here Jun 09 16:29:07 I think libopkg is no longer used in that way, but there is code for callbacks/progress still Jun 09 16:30:14 adelcast: no palmtops anymore Jun 09 16:30:25 android took over Jun 09 16:30:52 ah, I see, so OpenMoko had a UI on top of opkg? Jun 09 16:31:05 I can dig in git history Jun 09 16:31:07 yep Jun 09 16:32:26 openmoko 2007.2 was huge rewrite of ui. I do not remember did we had packagemanager ui at that time Jun 09 16:32:55 https://marcin.juszkiewicz.com.pl/2008/03/17/good-bye-ipkg-welcome-opkg/ :D Jun 09 16:34:05 ah, great!, thanks! Jun 09 16:35:15 so was OpenMoko -> OpenEmbedded Jun 09 16:35:28 makes sense Jun 09 16:36:34 openmoko vanished at some moment and we were main user of opkg so we adopted it Jun 09 16:36:41 where we == OE devs Jun 09 16:37:27 do you happen to know history behind opkg on OpenWRT? AFAIK OpenWRT is using a fork on a random commit of OE opkg, then they cherry-pick Jun 09 16:37:33 adelcast: main package format in OE was ipk since beginning. so nothing strange that OE took care Jun 09 16:37:50 I only use openwrt. never looked at their history Jun 09 16:38:38 I really tried to have them use our fork, merged a lot of their patches, but they never took the next step... Jun 09 16:38:56 'our fork'? Jun 09 16:39:11 sorry, not fork, our opkg Jun 09 16:39:35 they probably prefer to stay with known-to-work one Jun 09 16:39:46 (was still thinking of the initial ipkg to opkg fork) Jun 09 16:39:51 hrw: bittersweet flashbacks to thomas working on opkg at OpenedHand Jun 09 16:40:02 rburton: o yes Jun 09 16:40:18 he was main dev of opkg for a while iirc Jun 09 16:40:30 dpkg+apt was just too heavy Jun 09 16:41:45 which reminded me https://marcin.juszkiewicz.com.pl/2008/10/24/merging-stuff-from-poky-into-openembedded/ when I was digging in Poky source and merging lot of stuff back to OE Jun 09 16:41:54 I just did a core-image-minimal with ipk, deb and rpm (with package-management) Jun 09 16:41:57 "I have over 3 thousands of revisions exported (using git format-patch) from Poky and I am reviewing them and adapt to add into OE." Jun 09 16:42:16 adelcast: remember that in 2008 64MB ram/flash was still popular Jun 09 16:42:25 ipk: 4.6 MB, deb: 37 MB, rpm: 245 MB Jun 09 16:42:31 for rootfs ext4 Jun 09 16:42:58 definitely something to look at Jun 09 16:43:07 as numbers look too weird Jun 09 16:43:25 is perl for deb and python for rpm Jun 09 16:43:42 (dnf is python based, dpkg needs perl for its scripts) Jun 09 16:43:42 python for yum/dnf? Jun 09 16:43:48 minidnf is C Jun 09 16:44:05 and should work fine Jun 09 16:44:09 patches welcome! :) Jun 09 16:44:53 =) yeah Jun 09 16:44:58 rburton: I once did 'let me update rpm support' series. 78 patches sent to ML Jun 09 16:45:20 first few removed every trace of rpm support and then next ones added rpm support from Poky Jun 09 16:45:37 ha Jun 09 16:46:02 I remember 'you think that anyone will review it on ML??????' from some people Jun 09 16:46:12 git repo link was in first mail Jun 09 16:47:28 heh, old times Jun 09 16:47:51 when some software had 'blame hrw' releases Jun 09 16:48:36 OPIE 1.2.2 was 'fix UI on >QVGA screens ffs' with few changes done by people and lot of done by me Jun 09 16:49:05 Poky 3.1.2 was 'pinky' release because I had a pile of fixes to get it working on updated hosts Jun 09 16:49:08 etc Jun 09 16:49:45 thanks a lot guys, this helps a lot Jun 09 16:50:30 yw Jun 09 16:51:14 adelcast: would be nice if you share slides and video of presentation here after conference Jun 09 16:51:36 sure thing, I can do that Jun 09 16:53:08 eh, years since I did something around OE Jun 09 16:53:16 was good to refresh memory a bit Jun 09 16:53:26 thanks adelcast Jun 09 16:57:31 * hrw off Jun 09 18:18:59 adelcast: I picked up opkg in 2013, maintained it for about 2 years before I handed it off to you Jun 09 18:19:13 I remember trying to talk to openwrt folks but they had major NIH syndrome Jun 09 18:19:43 They stayed on a commit from around 2011 or 2012 instead of updating Jun 09 18:20:09 They still have some of the horrible logic bugs I fixed up like not checking for errors during package extraction Jun 09 18:26:48 adelcast: https://mirror.cyberbits.eu/fosdem/2014/UB2252A_Lameere/Sunday/Underwater_Acoustics_to_Opkg.webm from 1h00 to 1h15 Jun 09 18:27:16 The sound quality is awful but you may find some useful info in that section of my talk about opkg Jun 09 19:03:51 thanks! ^paulbarker Jun 09 19:03:55 I'll check it out Jun 09 19:04:28 around 2016, there was an effort to merge all the patches needed by OpenWRT (by Florian I think?) Jun 09 19:04:53 but the OpenWRT community didn't switch...even if all the work was already done Jun 09 19:05:02 now they diverge again....which is a shame Jun 09 19:05:23 like this: https://arstechnica.com/information-technology/2020/03/openwrt-is-vulnerable-to-attacks-that-execute-malicious-code/ Jun 09 19:05:29 hi i'm new to embedded development. i am trying to understand serial communication in hardware level but i'm confused Jun 09 19:05:30 that's not a problem with OE opkg Jun 09 19:05:40 i looked at some serial interfaces such as RS-232, RS-422 Jun 09 19:05:50 i have a question about number of wire. if two devices are connected to each other with two wires(Rx wire and Tx wire), then we can say this connection is full-duplex ? Jun 09 19:06:00 i try to understand the relationsheep between wires and transmission mode(simplex, half-duplex, full-duplex) Jun 09 19:06:13 but now I guess opkg gets a black eye, even if its not OE's opkg, but the old/messy OpenWRT fork Jun 09 19:58:44 adelcast: I can't believe they still have https://git.openwrt.org/?p=project/opkg-lede.git;a=blob;f=libbb/unarchive.c;h=31486e2d1664f7207ba0cd9e3b44c929b937c80c;hb=HEAD#l316 Jun 09 19:59:44 Esch3r: That's not an OpenEmbedded question, there are probably better places to find answers elsewhere Jun 09 20:30:04 Esch3r: its too late now to properly explain for me, but if you get back tomorrow during EU office times i can give you a proper rundown. Jun 09 20:42:28 ^paulbarker: that is pretty terrible **** ENDING LOGGING AT Wed Jun 10 02:59:57 2020