<Romster>
why does it not exit if it needs to follow -L and -L is not specified
<Romster>
hate github downloads -_-
<beerman>
this might sound funny but i don't follow 😄 what do you mean exactly?
<Romster>
pkgmk hangs at the download from github and it doesn't time out. sits there indeffently.
<Romster>
their is a redirect to the tags/$version that does work. but of course it wont have the filename unless you either use content-disposition header or specify the filename yourself.
<Romster>
i've played with content-disposition and it is so unreliable to trust using that.
<Romster>
TL;DR; pkgmk uses curl by default and does not follow -L to the actual download and hangs
<Romster>
when it does exit from the command line curl invocation it gives a 0 no error exit code.
<beerman>
we do have a renames feature in place now
<Romster>
yes i said that in one of the previous sentances &
<Romster>
^
<beerman>
its early
<beerman>
¯\_(ツ)_/¯
<beerman>
not enough coffee yet
<Romster>
<Romster> their is a redirect to the tags/$version that does work. but of course it wont have the filename unless you either use content-disposition header or specify the filename yourself.
<Romster>
or specify the filename yourself. ie the renames=()
<Romster>
should we add -L to pkgmk curl or redo github Pkgfiles to use the rename.
<Romster>
sorry if i wasn't clear.
<beerman>
uhm, for me it works like it always did
<beerman>
long term changing to renames should be the best thing to do, imho
<Romster>
wish their tags also included the archive name. that would resolve it transparantly
<Romster>
i forgot to add -o name -_- unlike wget it needs to be told to write to a file.
<Romster>
still pkgmk hanged. maybe i got a problem in my pkgmk.conf
<beerman>
maybe
<Romster>
ah shitty download mirror -_-
<Romster>
and of course it doesn't say what url it is trying to get the file from. think it needs to show the mirror it's grabbing out of pkgmk.conf on the console
<Romster>
i'll look at adding a patch for that later.
<Romster>
i had a mirror set as kernel.org was hanging on downloads from any download command and browser a day or two ago.
<Romster>
and of course i forgot.
<Romster>
pebcak
<Romster>
curl was in it's connection and wait phase, i'll need to add a timeout on it. 30 seconds should be more than enough
<Romster>
s/connection/connect/
<beerman>
:q
<Romster>
pushed qt6
<beerman>
neat, thanks
ukky has quit [Quit: Adding HDD to install CRUX on main system]
<beerman>
much more fun than mesa-demos is python, imo. turns out i had a fundamental flaw with the dependencies for flit/poetry and who knows what, maybe these alternate pip install routines aren't even needed.
<beerman>
big doh moment... anyway 😄
<farkuhar>
it's puzzling to me that the qtwebengine footprint hasn't changed since Romster's commit bfed937a19, but the Pkgfile has gone back and forth since then on the question of whether to use the system icu or the bundled icu.
<farkuhar>
darfo helpfully pointed out in #crux the conditions for /usr/share/qt5/resources/icudtl.dat appearing in the footprint, and why some users are getting footprint mismatches.
<beerman>
yeah well.. it will build its own when its not installed, so it wants it installed, but people get pissed when you depend on it for some reasons.. imo in a case like that it should hard depend on icu and just call it a day 🤷♂️