jbk, thank you

i run illumos-gate from github and openindiana

and i'm never sure which is part of which :) illumos-gate includes much that i wouldn't think would be part of base

jclulow: is it possible to whack src.illumos.org upside the head? I'm trying to look at procfs stuff and I know, e.g. 14788 doesn't seem to be there.

[illumos-gate] 14985 adjtime could be more const -- Patrick Mooney <pmooney⊙pc>

danmcd: Should be up-to-date now

Thank you!

[illumos-gate] 14519 zfs_root should cache vnode -- Patrick Mooney <pmooney⊙pc>

jclulow: if it runs pretty fast now, can it be automated again?

I am giving... while sleep $(( 60 * 60 )); do ptime -m /opt/opengrok/bin/index ; done

a try

:P

We'll see how it goes

do we have a preference on #pragma pack(1) vs. __packed for a struct? IIRC the former came from studio before it had gcc attribute compatibility

I would prefer the attributes that all the compilers support, rather than the pragmas we patch in

(I'm not sure which ones we patch, and which exist just for us upstream)

__packed expands to __attribute((__packed__)) so that's probably the more widely used one

oh, it's definitely more widely used.

they say to avoid pragmas....

btw, do we have description of 6208626 ?

as noted here: src.illumos.org/source/xref/illumos…_data.c?r=7c478bd9&mo=1654&fi=54#40

well, we don't have 1.12 probably

I'll check the bug in just a sec

we do, but I don't remember how to use the bug thingy we have

oh wow, I added a URL field so I stopped forgetting

and then pasted that url field into a gist, because I'm smart

does fenix? know about 6208626

one for the fenix keepers, I suppose

:) clang is telling: ../port/fp/fp_data.c:54:7: error: unknown type name '_h_val'

__packed is definitely the way to go

tsoome: looks like you also need to pacify this check: _XOPEN_SOURCE - 0 < 600 (iso/math_iso.h)