[Bug 287430] devel/qt6-tools: fail to package: qdoc:No such file or directory

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=287430

--- Comment #8 from Chad Jacob Milios <milios at ccsys.com> ---
(In reply to Jason E. Hale from comment #6)

TL;DR: sure does! good to close, thanks!
maybe ${:!${LLVM_CONFIG} --version!:R} tho, no?

--More--(33%)

Retested with llvm20 in place and llvm19 selected. Thank you so much for the
excellent [and prompt!] work going that extra mile rather than telling us to
punt it to poudriere

--More--(END)
`cat $MUSINGS`

Nothing at all wrong with how youve implemented it but for your consideration
(for this or the next thing) since clearly u r a bona fide wizard of ports &
bmake:

1. ${:!foo!} [vs] `foo` (modifiers can chain, lazy eval vs late)
2. ${LLVM_CONFIG} --version [vs] ${MAKE} -C $elsewhere -VFOO (maybe sparse
tree)
3. :S/./ /g:[1..2]:ts. [vs] :R (if paranoid of !~ x.y.z in the
future/multiverse)
4. for good measure an explicit :tw (are semantics of :S/foo/ / very well
defined?)

so all chained together its:
${:!${LLVM_CONFIG} --version!:S/./ /g:tw:[1..2]:ts.}

Surely im overthinking it but i'm glad to discuss the pitch black magic with a
fellow o.g. :) iron sharpens iron. So then, might you happen to know a more
clear/concise method than the line noise that is my suggestion 3 yet which
makes fewer assumptions than :R must? (i mean an elegant one i'm overlooking;
obvs we can contrive a dozen various uses of :C///) This sorta
$1.split("$2").splice(${3:-0},${4:-HUGE_VAL}).join("${5:-$2}") situation is
pretty common, no?

Just 1 & 2, given our manageable expectations are pretty reliably met:
${:!${LLVM_CONFIG} --version!:R}

-- 
You are receiving this mail because:
You are the assignee for the bug.