That was I think easily fixed:

662a663
>   $w = 0;

but I also get this, which I'm not so clear how to fix:

Negative repeat count does nothing at /usr/lib64/perl5/vendor_perl/Cyrus/IMAP/Shell.pm line 688

This is line 688:

for ($c = 0; $c < @l; $c += int((@l + $n - 1) / $n)) {

In the absence of comments, I couldn't tell why the code needed to be so complex - based on observed behavior, I "rewrote" the offending code segment like this:

  for (@l) {
        $lfh->[1]->print("$_","\n");
  }
  0;

diff is here:

687,693c688,690
<   for ($l = 0; $l < int((@l + $n - 1) / $n); $l++) {
<     for ($c = 0; $c < @l; $c += int((@l + $n - 1) / $n)) {
<       if ($l + $c < @l) {
<         $lfh->[1]->print($l[$l + $c], ' ' x ($w + 1 - length($l[$l + $c])));
<       }
<     }
<     $lfh->[1]->print("\n");
---
> 
>   for (@l) {
> 	$lfh->[1]->print("$_","\n");

Hopefully someone smarter than me can figure out what I might or might not have done wrong. I tested "lm user/foo*" and "lm -subscribed user/foo" when logged in as either cyrus or foobar and got the same results before/after my "patch"

I think the original code was trying to format the output into multiple columns if the terminal was wide enough and the values narrow enough.

$w looks like "the width of the widest value", $n looks like "how many columns of that width we can fit", @l looks like the list of values

And then in the double loop, the outer loop is each line of output, the inner loop is each column within that line, and all the stuffing about with $l, $c, and $n is to retrieve and print the right element from @l at the right time, with an appropriate amount of spaces following it so that the next column aligns correctly.

I can't tell at a glance whether the column display sorts downwards first or across first, some comments in here sure would be nice, wouldn't they...

I wonder if anyone ever sees the output in columns, or if real mailboxes are usually wide enough that it never happens. I guess if someone has a big monitor and runs a terminal on it full screen...

The column formatting is cute, but I'm not convinced it's necessary. It's also hard to read -- at a glance I don't see a likely cause of the "Negative repeat count" warning, nor is the display order obvious -- and hard to read is the same as hard to fix.

Your patch looks like it solves the problems in the complex column processing by just always outputting a single column instead, am I right? That seems reasonable to me. Do you want to submit this as a PR?

I think all the calculation of $w, $ll, $n, $c are unnecessary once the column formatting is discarded, so you should also remove the parts that calculate those values. $l and @l look like they're still needed though.

I could reproduce this easily with:

localhost> cm user.12345678901234567890123456789012345678901234567890123456789
localhost> lm
Negative repeat count does nothing at /usr/lib64/perl5/vendor_perl/Cyrus/IMAP/Shell.pm line 689.
user.12345678901234567890123456789012345678901234567890123456789 (\HasNoChildren)
localhost>

interestingly, if you create the mailbox one byte shorter, it doesn't print this msg:

localhost> cm user.1234567890123456789012345678901234567890123456789012345678 
localhost> lm       
user.1234567890123456789012345678901234567890123456789012345678 (\HasNoChildren)
localhost>

The Use of uninitialized value $w was fixed by 42a10f4.