heh

just as a sanity check -- for mac_ether_offload_info_t the size of the 'data' portion of the packet (bytes after TCP/UDP header) would be meoi_len - sum(meoi_l[234]_hlen), right? (assuming the respective MEIO_L[234]INFO_SET flags are set)

(there's apparently an azure bug w/ UDP checksum offload that requires knowing this to avoid

jbk: that sounds correct

ok.. i figured.. but always hate to assume and then find out i'm wrong the hard way :)

richlowe: file another issue? :)

hrm though.. i suppose by the time we get the packet for TX, it's too late to say 'hey, we can't do it on this packet' isn't it?

gwr: #18194 (fenix?)

BUG 18194: test-runner default output directory doesn't work (New)

↳ illumos.org/issues/18194

heh.. test-runner feels like the gift that keeps on giving :)

jbk: it's been great for years and years, and it's notably not either of Sun's frameworks...

though I think STF is in there somewhere for zfs

my favourite is still dtest, because absolutely as simple as it possibly can be

Unfortunate name. I see it as "detest" which is clearly a wrong feeling for something helpful.

it's the dtrace test runner. The pun might be intentional, they're punny people

and it's not a simpson's reference, so that only helps the chances of it being something else

or 'dee (the) test'

unlike my illumos branch named 'theperlthe' :)

(speaking of simpsons references)

You touch it you own it :)

they still don't beat the linker aliens for getting puns into /usr/bin, with crle, moe, & lari

gwr: not so much that, as every time I merge with you I find something new. The vast majority have been my fault

I had let enough false "32-bit" tests through it actually makes a noticeable different to runtime too :\