ya, fortunately fusion has gdb stub and we can inspect the machine:)

[illumos-gate] 16820 Ignore library/libxml2/32 during package lint -- Marcel Telka <marcel⊙ts>

ho

[illumos-gate] 16670 libshare: issubdir() error: strcpy() 'dir1' too large for 'childdir' -- Toomas Soome <tsoome⊙mc>

tsoome_: if you're using source debugging that way, have you seen arm64-gate's .gdbinit?

it's not perfect (we don't have a way to find the .bss again for it, and it doesn't care about alignment), but it works like, many times out of 10

unfortunately, the one time it doesn't will really confuse you :\

anyway, that's what I did before I ported kmdb, and now I often use both at the same time (because I ported kmdb and don't trust it fully yet)

I usually do not use source debugging with gdb stub, I have done this just few times:) but it is good to know, thanks

tsoome_: you're gonna hate and/or love open-std.org/jtc1/sc22/wg14/www/docs/n3356.htm

well, as it correctly stated, if we can declare variable in for statement, then why stop there -- it does make sense

I was just remembering how you dislike if ((foo = bar()) == NULL)

which isn't that different from if (foo = bar; foo != NULL)

which seems allowed here regardless of whether the LHS is a declaration

when I say "isn't that different", you're meant to ignore I reversed the condition like an absolute crumpet

well, I still dislike excessive use of assignment statement in if:D but if you add variable declaration to it, that is different;)

I could grow to like that. I'm always trying to scope things down as tightly as possible.

My preference for assignment in if is definitely at odds with my preference for scope minimisation in C haha

Hard to reconcile sometimes

I just dont like _statements_, really.

I certainly like (declaration; value check) better than combining them into a single ((foo = bar()) == NULL) style combo value-setting+check