lavaball has quit [Remote host closed the connection]
lavaball has joined #crux-social
zorz has joined #crux-social
<remiliascarlet>
SiFuh: It's as if the Trump haters want him to win.
<remiliascarlet>
lavaball: Everyone agrees that Nintendo Online sucks. The only thing they added to it is a paywall. Still remember all the Nintencucks saying how paywalling a service that used to be free of charge for 2 decades was somehow a good thing, because it would "improve the experience". No improvement at that front at all.
<remiliascarlet>
But then again, the 2 main businesses Nintendo is involved in is 1) game development, and 2) suing their own fans. Network infrastructure has never been their strong point.
<lavaball>
remiliascarlet, you could help me with something. look for Table 1 Rates of Reported Complications in https://www.fda.gov/media/75792/download?attachment and explain to me what in the death row 0.5 means.
<lavaball>
did half a person die?
<remiliascarlet>
Possibly like Youmu, who's half human, half ghost.
<lavaball>
seriously, take a look for me, if you have the time.
<lavaball>
i doubt that has anything to do with the link i gave you.
<lavaball>
says Cases/Million Vaccinations. so 0.5 that would be 5000 people? did 5000 people die?
<lavaball>
remiliascarlet
<lavaball>
?
<remiliascarlet>
I don't know, I have more important things to do at the moment.
<remiliascarlet>
More important than figuring out what the government means by "0.5".
<lavaball>
did you look at the table?
<ppetrov^>
lavaball, in the Table 1 legend it says "Cases/Million Vaccinations"
<ppetrov^>
so 0.5 per 1mil
<ppetrov^>
get the total number and you can calculate how many people this 0.5 refers to. So, for example this would imply that 1 in 2,000,000 people died
<ppetrov^>
1:2,000,000
<ppetrov^>
similar to remiliascarlet i have other stuff to do at the moment
<lavaball>
great. i wanted to know if remiliascarlet understands it.
<farkuhar>
I only started digging into their git repo after hitting iproute2 build errors on CRUX-MUSL. Between versions 6.10 and 6.11 they introduced new features, which rely on glibc-specific headers to compile cleanly.
<farkuhar>
The source being cited for these man-page revisions has three tiers of word lists. "Sanity check" is in the middle tier, so according to them lavaball isn't committing the worst offense with his repeated use of "dementia boy".
<farkuhar>
As for the build errors on CRUX-MUSL, one of them is easy to fix by inserting "#include <endian.h>" at the top of libnetlink.h, but trying to compile the new file bridge/mst.c produces a more annoying error: incompatible pointer type.
<lavaball>
farkuhar, do you have your 5 brain cells together? i can foreshadow what i'm actually going to do them.
<farkuhar>
Now the incompatible pointer type error was preceded by complaints about redefined structs under /usr/include/netinet/. I'm guessing it's those redefinitions that were responsible for changing the type of the pointer passed to a function in json_print.h
<farkuhar>
So musl and glibc must be populating /usr/include/netinet/ with headers that disagree in their definitions of the relevant structs.
<farkuhar>
I'd prefer not to unpack glibc's /usr/include/ to run a diff against musl's netinet subdirectory, but if that's the only way to debug the iproute2 build errors, that's what I'll have to do.
lavaball has quit [Ping timeout: 260 seconds]
lavaball has joined #crux-social
<ukky>
farkuhar: can you upload build log of 6.11 without any new patches?
<farkuhar>
ukky: are you sure you want the build log without even the trivial addition of "#include <endian.h>"? It's not very informative when the build stops at an implicit declaration error, rather than proceeding to the file affected by missing #defines.
<ukky>
yes, I'd like to see original build failure.
<farkuhar>
The inlined functions in the iproute2 source tree are hidden behind #ifdef tests, so they aren't being activated on glibc-based systems. Meanwhile, musl has a different set of #defines than glibc, so the inlined functions are clashing with what was already defined in the various headers. Hence the incompatible pointer type error.
<ukky>
HAVE_SETNS is not properly defined for musl, thus setns() gets redefined
<farkuhar>
I got past that error by inserting "#define HAVE_SETNS" into one of the files, but I haven't tracked down the relevant #define that fixes the compilation of bridge/mst.c
lavaball has quit [Quit: lavaball]
<ukky>
farkuhar: could you post the log when you build with endian.h?
<farkuhar>
Hahaha "fatal error: ... no space left on device" <- sounds like something zorz would see after allocating all that space for his distro-hopping adventures. In my case it's because I'm working at an underpowered netbook, while my desktop is building nodejs.
<ukky>
could this be a reason for configure failure?
* farkuhar
runs ncdu to see what's taking up all that space on the root partition
<farkuhar>
The test code in the configure script compiles cleanly now. I'll try running the entire configure script again, to see if it still reports 'libc has name_to_handle_at: no'
<farkuhar>
Okay, configure is correctly detecting musl's implementation of name_to_handle_at(), but the build still stops at the new file bridge/mst.c (same as happened on the desktop machine, which didn't run out of disk space).
<remiliascarlet>
farkuhar: So it looks like iproute2 is facing a similar problem as GNU binutils has been for the last couple of versions. I had to pick 2.37, because 2.43 wouldn't compile.
<remiliascarlet>
Except it's not really a musl problem, I get the exact same problem on glibc, and tested across multiple distro's on both glibc and musl. Any version from 2.38 is simply not compilable.
<farkuhar>
remiliascarlet: Are the GNU binutils devs also becoming woke, and updating their man-pages with "more inclusive" language? If that's what they prefer to spend their time on, we shouldn't be surprised that compilability is suffering.
<remiliascarlet>
farkuhar: I didn't check the manpage for that. I wonder if they'd rename "manpage" to "personpage" to be "more inclusive"?
<remiliascarlet>
After all, the GNU guys tend to be not only woke, but autistic too.
<farkuhar>
remiliascarlet: I didn't check the manpages directly, I just read the commit messages in the iproute2 git repo. (A git repo I would have no reason to visit, if the latest version of iproute2 had actually compiled.)
<remiliascarlet>
But really, in what way are they supposed to "inclusivy" the manpage of iproute2 of all programs?
<farkuhar>
The three latest commits to the man-pages indicate that some "non-inclusive" language was removed: whitelist, segregate, and sanity-check.
<remiliascarlet>
"sanity-check" Good lord, at least they seem to be aware that they're fucking insane...
<ukky>
farkuhar: glibc does not declare in6_addr etc upon some conditions, i.e. when user has included linux network headers. But musl declares in6_addr unconditionally.
<farkuhar>
Actually "sanity-check" only appeared as a comment in the source code, not in a man-page, so most downstream users would never see it.
<remiliascarlet>
I mean, the way they see "sanity-check" as "non-inclusive".
<farkuhar>
ukky: thanks for investigating. Any ideas how we could fix the build for CRUX-MUSL?
SiFuh has quit [Remote host closed the connection]
SiFuh has joined #crux-social
<ukky>
farkuhar: I will try to compile with i686-musl, as my amd64-musl system currently runs netbsd. If that fails, I will try chrooted environmwent.
lavaball has joined #crux-social
zorz has quit [Quit: leaving]
lavaball has quit [Remote host closed the connection]