Re: *more* relocatable perl
[email protected] (Darren Duncan) Sat, 18 Jul 2026 23:29:45 -0700
| Newsgroups | perl.perl5.porters |
|---|---|
| Message-ID | <[email protected]> |
On 2026-07-18 4:13 p.m., Dan wrote: > On Sat, Jul 18, 2026 at 6:36 PM Darren Duncan wrote: > > On 2026-07-18 2:15 p.m., Gianni Ceccarelli via perl5-porters wrote: > > On 2026-07-18 Darren Duncan wrote: > >> As long as the user knows where the "main program" is and invokes > >> that, it can find all of its other pieces and dependencies relative > >> to its own location implicitly without having to be fed any extra > >> parameters and without any recompilation > > > > Yep! Exactly that. > > > > And a perl built with `userelocatableinc` is *very close* to that, > > except for a handful of %Config entries. > That leaves me to question why the behavior of userelocatableinc is not the > default, what are its downsides that it isn't simply how perl always works. -- > Darren Duncan > > A necessity of being relocatable is relative paths, which are a security nightmare. That's a good reason to not be default. But is this a problem just specific to Perl or to programming language interpreters, or do all the other kinds of applications, typically for end users, that are typically relocatable, just not consider the security nightmare to be a problem? Darren Duncan