dangole has quit [Remote host closed the connection]
rua has quit [Read error: Connection reset by peer]
rua has joined #openwrt-devel
foxtrot has quit [Quit: Goodbye]
slh64 has joined #openwrt-devel
foxtrot has joined #openwrt-devel
<nbd>
hauke: i wasn't active looking into tls 1.3, i just worked with uclient on my host system for testing and noticed ustream-ssl connection failures
rz has quit [Ping timeout: 480 seconds]
rua has quit [Quit: Leaving.]
robimarko has joined #openwrt-devel
<f00b4r0>
Mangix: 1) this is because we're doing fancy color stuff (dispensable), 2) make 4.1 was released in... 2014, we started requiring it in... 2021. There is a difference between "updating to a tool several years old" and "updating to a tool released a few weeks/months ago". Furthermore make does not require an entire interpreter environment (python) to even execute. I rest my case.
<ynezz>
Mangix: robimarko: Re: git.openwrt.org/e09d3fd290 - BTW my understanding of that previous comment "# make core tools depend on sed and flock" and code `$(filter-out zstd,$(tools-core))` was, that you actually want sed/flock dependency for all the tools-core, but not for zstd
<robimarko>
ynezz: Comment is a bit misleading
<robimarko>
Because this commit doesnt really change anything
<robimarko>
That for loop was making all core tools depend on sed, and then flock was made a dependency of sed
<robimarko>
Nothing has really changed, flock is still built first as sed depends on t
<robimarko>
Then libdeflate, zstd and sed
<ynezz>
ok, that description got me, still doesn't explain the build failure, going to nuke the build dir and check again
<robimarko>
ynezz: do you have a log?
<ynezz>
nope, I've a gnome-terminal which crashes on middle-click :P need to update
<ynezz>
it might not be even related, but I've checked whats new in the tree and spotted this
<ynezz>
tools build now fine, so not related apparently, sorry for the noise and thanks for making xz usage reduced, really appreciate it
<robimarko>
Relief that I did not break more stuff, still weeding out all of the hashes that need to be updated
<ynezz>
its a huge change, so some fallout is expected, I wouldn't worry about that :P
<robimarko>
Anybody here merging packages repo stuff?
<robimarko>
It seems that I managed to merge the PR to GH only, its not mirrored to git.openwrt.org
<robimarko>
Well, all toolchain dockers should have been updated by now
<robimarko>
I see at91 have been pushed 19 hours ago
<f00b4r0>
it only happens for one arch though
<robimarko>
it is at91 specific package
<f00b4r0>
I just noticed after downloading the artifact
<f00b4r0>
and it's also doing funky things, it builds either of two different versions (with different hashes) depending on config option, if I read this right?
<f00b4r0>
the Makefile was refreshed 7 months ago so presumably it was missed?
<f00b4r0>
it's pulling source from GH
<robimarko>
oh yeah, it changes the hash it pulls based on config option
<robimarko>
It hasnt been touched outside of the CPE identifier for 2 years
<robimarko>
and even then ci did not actually build it since it only built for malta and x86
<robimarko>
I refreshed the "main" hash but CI is building for sam9x so that changes the hash being pulled
<robimarko>
Yeah, when building for sam9x then hash is incorrect
<ynezz>
f00b4r0: thanks for letting me know, I'll change the pattern to *.tar.*
<robimarko>
I am going to push this as follow-up fix to not trigger CI rebuild
<f00b4r0>
ynezz: they're all formatting + path fixes? if it works i see nothing wrong (I can't test anyway), though I'm not sure splitting string literals at sometimes odd places really helps readability. It also makes manually testing the quoted commands much harder, which is a downside imo.