I tried to fix this, and it didn't seem like it was going to be possible

richlowe: I don't think it's impossible, but it would be a royal PITA to do it.

compat versions of the functions that touch "struct tm" get the old external names (to keep the ABI intact); new ones get revised external names for linkage, and never let the streams cross..

"struct tm" changing between old ABI and new ABI is your biggest PITA, but with that model I *think* it's doable. See the large-files examples from times past, I think.

if there are other structs which have embedded "struct tm" it gets hairier.

my recollection is that largefile isn't a great example, because FILE is (just about) opaque, and off_t is integral

but it's been a while

making the basics work along sommerfeld's lines worked, but actual software -- especially if not all rebuilt together -- was hairy

if you a 'foo' with a new tm, but a libfoo with an old-tm.

Yes. The problem is passing a foo between things. Particularly an old one to a new thing.

things like this are much easier to fix when you first establish a new ABI, like amd64 or aarch64

like FILE becoming opaque in the Solaris 64-bit ABI to avoid the mistakes of the 32-bit ABI

alanc: believe me, I have thought about it, but then we have "it only works on aarch64"

Yeah I wouldn't want to make that sort of change just on one architecture

I think it would do more harm than good with downstream software

I _do_ think we could specifically address this in the manual page with example code though

like, "oh, you want tm_gmtoff, sorry, do this...'

I actually wonder if you could do something _utterly horrendous_ with a _macsro_

*macro

probably not, I guess

adding _apologia_ to our manual pages is a fine tradition, _apologies_ might be new though.

I think time.h(3HEAD) should also mention the things that are missing and redirect to how to get them

[illumos-gate] 17231 DMU: Fix lock leak on dbuf_hold() error -- Alexander Motin <mav⊙Fo>

[illumos-gate] 16796 loader: allow fs modules to use ioctl -- Toomas Soome <tsoome⊙mc>

[illumos-gate] 16797 loader: dosfs should be able to work with different sector sizes -- Toomas Soome <tsoome⊙mc>

[illumos-gate] 17280 Tidy up grpck references in the manual -- Peter Tribble <peter.tribble⊙gc>

danmcd I think it ws you that mentioned it on mastodon a while ago, did you get one of those PCIe Gen 4 x1 10Gbit ethernet adaptors?

If so, was that one that worked on illumos?

[illumos-gate] 17198 stream_runservice() eagerly helps us deadlock -- Andy Fiddaman <illumos⊙fn>

I only have 1x PCIe Gen 4 x1 left so can't fit my usual Intel X510 DA-1

IIRC it's not supported and it's an AQC113 or similar.

Ah yes that model nummer seems vaguely familier

OK, no point in grabbing one if it's not supported.

unless you have the time to add support!

*sweats* I have 2 failed attempts of adding a CH320 serial support

I don't think a NIC will fair much better

richlowe: I'm boring and stick notes like that under Usage instead of trying to teach people how to spell Apologia

USAGE

This command is primarily used by the Service Management Framework

(SMF). For an equivalent to the killall command found on other systems

that allows specifying criteria to determine which processes to signal,

see pkill(1).

do any of you use pcitool much, and want to confirm what it means by "bank" in the usage?

or a general usage of `-n`

it doesn't help when a command tells me it already knows it might panic the system, when I'm trying to stop it panicking the system