Notices by Luke T. Shumaker (lukeshu@fosstodon.org)

My apologies to WIZnet, it seems my SPI is buggy and is right-shifting reads by 1 bit. 0x11<<1 is indeed the valid SOCKMODE_UDP.

And the reference implementation (ioLibrary_Driver) has an IPRAW mode, but no mention of that in the datasheet...

The W5500's datasheet is even more incomplete than I thought. I've done something that makes the chip say Sn_SR=0x11, which the datasheet doesn't list as a possible value.

But I can't figure out how to get a wrench on those bolts without taking the catalytic converter off.

No really, the manual's step-by-step jumps straight from separating the knuckle to just "Remove right drive shaft from transaxle." (it calls the CV axles "drive shafts"). No mention that the right drive shaft has some more bolts holding it in place!

And I've watched a half-dozen YT videos of people showing how to replace the CV axles on this car. Everyone just shows the left side which doesn't have this step!

(Big thanks to this video of a '97 which is similar enough https://www.youtube.com/watch?v=zKSp2Ey_FTw )

Something I miss about the old #Identica is how weirdly well it ranked on Google. Like I could dent something like "The factory service manual doesn't mention it, but when removing the passenger-side CV axle on a 2000 #Infiniti #i30 you need to unbolt the support bearing (part 22 in the diagram) from the support bracket (part 23)" and somehow that dent would suddenly be in the top 5 results for "2000 infiniti i30 cv axle". On #Mastodon sharing such information feels like shouting into the void.

@brooke Is the reason Pico 8?

Specifically ARMv6-M. Get outta here with your fancy A-class processors.

why arm v6 have no NVIC_IABR register???

#define ceildiv(n, d) (( (n)+(d)-1 ) / (d)) /* ceil(n/d) */
#define floorlog2(x) (64-__builtin_clzll(x)-1) /* floor(log2(x)) */

#c #gcc

#define lambda(ret, args, body) ({ ret lambdafn args body lambdafn; })

#gcc #gnu #c

@foo Samsung sort of did that with "One Connect", except they made a proprietary plug for it. A little too evil capitalist I guess...

@mattl Non-recommendation: I have an OseTub 7-port switch (with on-device buttons and a remote).

While a "please wake the screen" packet from any input will wake the TV, it won't set that to be the active input. So then the TV will be on but just have a blue "no signal" screen until I hit the switch's button for the input that woke the TV.

Maybe also some weirdness when switching between inputs with different resolutions?But that might be the TV itself.

Hot take: #RFC 2131 and 2132 (DHCPv4) are unusually poorly written for RFCs. #ietf

(Of course it's not quite fair to pick on documents that are almost as old as I am and were written before many guidelines for RFCs existed.)

@lowqualityfacts

Me: *reads the post out loud*
Roommate: That's so dumb.
Me: Would you say that it's low quality?
Roommate: It is very low quality.

@jonathanmatthews @ctietze

(highest precedence)
$GIT_EDITOR
$(git config core.editor)
$VISUAL
$EDITOR
compile-time DEFAULT_EDITOR ("vi" by default, but your distro might change that)
(lowest precedence)

reference: `man git-var`

@evan Does SSH Communications Security count? (I'm not sure how they were funded in the early days.)

@lxo thanks for getting -fcallgraph-info landed all those years ago, it's very useful to me now!

(Funny how much I see your name in #GNU commit logs :) )

The #FreeSoftware movement is fundamentally a political movement advocating for folks' freedoms.

To say that politics have no place in the Free Software movement is to say that the Free Software movement shouldn't exist.

But it's an attractive line to those looking to exploit the #FOSS software that the movement produces while subverting the movement itself.

@lxo I meant that the kernel supports too many syscalls, not that too many syscalls are actively running :)