Re: [PATCH v2 storage 12/15] iscsi: add support for non-persistent discovery

[Mira Limbeck <m.limbeck@proxmox.com>]

On 5/23/26 11:28 PM, Thomas Lamprecht wrote:
> On Thu, 30 Apr 2026 19:27:10 +0200, Mira Limbeck wrote:
>> diff --git a/src/PVE/Storage/ISCSIPlugin.pm b/src/PVE/Storage/ISCSIPlugin.pm
>> @@ -226,7 +227,7 @@ sub iscsi_discovery {
>>          # In case of multipath we can stop after receiving targets from any available portal
>> -        last if scalar(keys %$res) > 0;
>> +        last if defined($target) && scalar(keys %$res) > 0;
> 
> $target is declared with `my` inside the outfunc closure a few lines up, so
> it is out of scope here - this doesn't compile under `use strict` ("Global
> symbol $target requires explicit package name"). 13/15 rewrites the line to
> use $target_found, so the final status is again fine, but this commit on its
> own breaks the build. Worth folding that fix into this commit for
> bisectability.
> 
> btw. I like using rebase + exec for ensuring that every commit is clean
> from code format and testing/build POV. It's slightly nicer in rust projects
> where just cargo fmt / check / test calls can be used in a generic fashion
> independent of the specific repo layout, but most of our perl repos are
> rather uniform in its basic targets nowadays, so something like:
> 
>  git rebase --update-refs -x 'printf "\n-------\n"' \
>    -x 'git show --no-patch --oneline --no-decorate HEAD' \
>    -x 'make tidy' \
>    -x 'make -C src test' \
>    origin/master
> 
> should do the trick (btw. --update-refs is normally not required, but
> essential for stacked branches for (sub) series, and it normally doesn't
> hurt to pass).
> We might want to add something like that to the developer wiki, or maybe
> even add a proper (e.g.) "dev-test" make target that combines the four -x
> steps (for me my shell history got me covered ^^).

Thank you for the tip!

I've rebased it so many times, but didn't think of doing an exec and
building every commit on its own.