Fantastic :) this is the answer I was waiting for ;) I will see if they can do the change in the code base, meanwhile I will go my own way patching. Thanks a lot, Gabriele ---------------------------------------------------------------------------------------- Sonicle S.r.l. : http://www.sonicle.com Music: http://www.gabrielebulfon.com Quantum Mechanics : http://www.cdbaby.com/cd/gabrielebulfon Da: Erik Trimble A: discuss@lists.illumos.org Data: 29 agosto 2016 10.06.48 CEST Oggetto: Re: [discuss] default system shell Your RedHat friend is, frankly, talking out of his ass. First off, Ksh (also known as ksh88) is fully, 100% backwards compatible with the Bourne shell (sh). Secondly, ksh93 is also 100% compatible with Bourne shell (sh), but is NOT 100% compatible with ksh88 (a couple of things changed). Thirdly, the POSIX shell standard specification is based on ksh88, with a couple of minor changes. The result is that ksh93 isn't quite absolutely 100% POSIX compliant. Bourne Shell is not POSIX-compliant at all. Fourthly, Bash started out as "enhanced" ksh93 (different codebase though), and theoretically was Bourne-compliant at that time. This was two decades ago, and Bash has long, long, long since given up being 100% Bourne compatible. Even with the --posix flag, it's not 100% Posix Shell compatible either. Here's a not-so-short list of differences (both features and behavior) of Bash vs Bourne shell: http://www.gnu.org/software/bash/manual/html_node/Major-Differences-From-The-Bourne-Shell.html The local variable problem is almost certainly a Bash-ism. That is, the code is using a Bash-specific feature, and committing the cardinal developer sin of assuming their local system setup is What the World Uses. Ksh (both ksh88 and ksh93) supports local variables, but the syntax is different. Bourne Shell doesn't support local variables. You should inform the upstream code maintainers that they have a major portability mistake, and that THEY should fix this. It would be nice to provide them with patches, but this is a problem that isn't just you, it's in the codebase itself. This kind of assumption is on the same level as assuming all modern CPUs are little-endian, just because most people run x86. If a developer is going to use features in a shell that assume a certain shell, then that shell should be explicitly invoked by the script, not presuming the system is configured the same way the developer's environment is. I get sensitive to this, because I'm a Systems person, and I see this kind of hubris all the time, and it causes an unending stream of headaches that inevitably get laid at my door, rather than the developer, who is the guilty party. -Erik illumos-discuss | Archives | Modify Your Subscription
------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com
