You are not logged in.
I have a custom repo hosted as a bucket on Amazon S3.
Turns out pacman -S is unable to download packages from S3 whose filename contains a +.
Example: `avahi-1:0.8+r127+g55d783d-1-x86_64.pkg.tar.zst`
It appears that pacman does not escape the pluses in the URL, while the S3 http frontend expects those to be escaped. Reading RFC 3986, pacman appears to be in the right here ("+" is in the list of "sub-delims" which are valid unescaped for "pchars" in "paths").
The download URL for the package as shown on aws.amazon.com/s3 ends is .../avahi-1%3A0.8%2Br127%2Bg55d783d-1-x86_64.pkg.tar.zst . Replacing the %3A with colon still works, but not the %2B.
Anybody else have that problem, or have any good ideas what to do here?
Last edited by jernst (2023-09-27 20:30:54)
Offline
https://man.archlinux.org/man/core/pacm … en#OPTIONS
You could use the XferCommand and encode the url before (jq is probably the most readable solution)
https://gist.github.com/cdown/1163649
https://gist.github.com/jaytaylor/5a90c … 847ce58736
http://sfyang-en.blogspot.com/2012/04/u … cript.html
Offline
Ha! I hadn't known about that option.
curl -o $2 $( echo $1 | sed -e 's!+!%2B!g')in a separate shell script seems to do the trick.
Offline
Please always remember to mark resolved threads by editing your initial posts subject - so others will know that there's no task left, but maybe a solution to find.
Thanks.
Offline