A POSIX and a UTF-8 question

A POSIX and a UTF-8 question

[Daniel Qarras]

Hi,

and pardon me but I've got few lame questions regarding the upcoming
zsh-4.4..

- is there any documentation explaining conformance of zsh as /bin/sh
with the latest POSIX sh spec?

- any ideas when zsh-4.4 supporting UTF-8 could be released? As a Linux
user I see that Debian and RHEL are being prepared for a release by the
end of the year and as they will be in use for years to come it would
be great if zsh with UTF-8 support would be available on them by
default.

Cheers!


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: A POSIX and a UTF-8 question

[Bart Schaefer]

On Sep 4,  1:24pm, Daniel Qarras wrote:
} 
} - is there any documentation explaining conformance of zsh as /bin/sh
} with the latest POSIX sh spec?

No, not really.  What exactly do you mean by "the latest" by the way?
I've been following their mailing list, and there's a new revision
coming up for publication later this year.
 
} - any ideas when zsh-4.4 supporting UTF-8 could be released? As a Linux
} user I see that Debian and RHEL are being prepared for a release by the
} end of the year and as they will be in use for years to come it would
} be great if zsh with UTF-8 support would be available on them by
} default.

My guess is that it's already too late for any newly-released version of
zsh to be considered stable enough to go into RHEL5.  I don't know about
Debian.  Clint?

Re: A POSIX and a UTF-8 question

[Clint Adams]

> My guess is that it's already too late for any newly-released version of
> zsh to be considered stable enough to go into RHEL5.  I don't know about
> Debian.  Clint?

We can probably squeeze a new version in as late as mid-October.

Re: A POSIX and a UTF-8 question

[Daniel Qarras]

Hi!

> } - is there any documentation explaining conformance of zsh as
> /bin/sh
> } with the latest POSIX sh spec?
> 
> No, not really.  What exactly do you mean by "the latest" by the way?
> I've been following their mailing list, and there's a new revision
> coming up for publication later this year.

I noticed this also and that's why I was talking about the "latest and
greatest" but of course it is more important to know conformance
regarding to any standard version, not just the latest.

Thanks!

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: A POSIX and a UTF-8 question

[Peter Stephenson]

On Tue, 5 Sep 2006 11:23:34 -0700 (PDT)
Daniel Qarras <dqarras@yahoo.com> wrote:
> > } - is there any documentation explaining conformance of zsh as
> > /bin/sh
> > } with the latest POSIX sh spec?
> > 
> > No, not really.  What exactly do you mean by "the latest" by the way?
> > I've been following their mailing list, and there's a new revision
> > coming up for publication later this year.
> 
> I noticed this also and that's why I was talking about the "latest and
> greatest" but of course it is more important to know conformance
> regarding to any standard version, not just the latest.

Unfortunately, the documentation is "if you start zsh in sh mode [how to
do this is in the normal documentation, at least] it sort of works most
of the time".  I'm sure there are many bits that aren't quite right, but
nobody's kept track of them all, and I don't suppose anybody's got the
time.  It's one of the disadvantages of having only a handful of us
actively maintaining the internals in our spare time.  This task
wouldn't need familiarity with the C code, though.

One recent headline feature is that in 4.3 (but not 4.2) Wayne has fixed
it so that (as far as I know) word splitting in sh mode is according to
other shells.  That has probably been the most noticeable problem in
recent years.

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

Re: A POSIX and a UTF-8 question

[Daniel Qarras]

Hi!

> One recent headline feature is that in 4.3 (but not 4.2) Wayne has
> fixed it so that (as far as I know) word splitting in sh mode is
> according to other shells.  That has probably been the most
noticeable
> problem in recent years.

I now actually tested on my FC5 laptop linking /bin/sh to a recent
zsh-4.3 from CVS and I was astonished how well it worked! I'm writing
this article from a system with /bin/sh being zsh-4.3!

There were only two glitches that I needed to fix by hand and after
that all services seem to go up and down without any problems:

1) There is unfortunately a bash-only expression used in RH's shell
function library, that is =~. However, this was used only once in
device mapper related function so that did not affect me, I just
commented it out. YMMV. (I did not find a www description for this but
it can be at least checked from bash3.1 manual pages what it does.

2) It seems that many RH scripts do something like:

cd /some/path
. functions

and in zsh-4.3 as /bin/sh this does not work, it needs to be:

cd /some/path
. ./functions

Additionally, there's one cosmetic bash feature used that seems to be
not supported by zsh: RH uses in their i18n effort tricks like:

echo -n $"Starting service foo: "

With bash I get just the normal string, with zsh I also get the $. I
did not look very carefully about this, so not sure what they are
actually doing here.

I filed a note to RH bugzilla about 1 and 2 as 206035 but in any case I
am very impressed how well zsh nowadays works as /bin/sh in a Linux
distribution.

Cheers!



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: A POSIX and a UTF-8 question

[Peter Stephenson]

Daniel Qarras wrote:
> 1) There is unfortunately a bash-only expression used in RH's shell
> function library, that is =~. However, this was used only once in
> device mapper related function so that did not affect me, I just
> commented it out. YMMV. (I did not find a www description for this but
> it can be at least checked from bash3.1 manual pages what it does.

That's for extended regular expressions.  That wouldn't be too hard to
support, if we used the system's RE capabilities, although we'd probably
want to do parentheses differently.  However, as you say, it's specific
to bash so isn't that vital.

> 2) It seems that many RH scripts do something like:
> 
> cd /some/path
> . functions
> 
> and in zsh-4.3 as /bin/sh this does not work, it needs to be:
> 
> cd /some/path
> . ./functions

Right.  I think zsh has always been implemented like that.  We don't
really want to change it; the standard (IEEE Std 1003.1, 2004 Edition)
says specifically:

    RATIONALE

    Some older implementations searched the current directory for the
    file, even if the value of PATH disallowed it. This behavior was
    omitted from this volume of IEEE Std 1003.1-2001 due to concerns
    about introducing the susceptibility to trojan horses that the user
    might be trying to avoid by leaving dot out of PATH .

I checked on my Fedora system (about Core 4, I think) and the only init
script without a path is network:

  cd /etc/sysconfig/network-scripts

  . network-functions

If it's still there it probably ought to be fixed.

> Additionally, there's one cosmetic bash feature used that seems to be
> not supported by zsh: RH uses in their i18n effort tricks like:
>
> echo -n $"Starting service foo: "

It would be relatively easy to interpret these in the same fashion as
some other form of quoting.  However, there's not much
internationalization in these parts and won't be without a huge effort
and it's not clear it's worth the effort doing anything apart from that.

Thanks for the notes.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


To access the latest news from CSR copy this link into a web browser:  http://www.csr.com/email_sig.php

Re: A POSIX and a UTF-8 question

[Daniel Qarras]

Hi!

> > 1) There is unfortunately a bash-only expression used in RH's shell
> > function library, that is =~. However, this was used only once in
> > device mapper related function so that did not affect me, I just
> > commented it out. YMMV. (I did not find a www description for this
> but
> > it can be at least checked from bash3.1 manual pages what it does.
> 
> That's for extended regular expressions.  That wouldn't be too hard
> to
> support, if we used the system's RE capabilities, although we'd
> probably
> want to do parentheses differently.  However, as you say, it's
> specific
> to bash so isn't that vital.

The bug I filed to RH is already handled. This will not change but...

> > 2) It seems that many RH scripts do something like:
> > 
> > cd /some/path
> > . functions
> > 
> > and in zsh-4.3 as /bin/sh this does not work, it needs to be:
> > 
> > cd /some/path
> > . ./functions
> 
> Right.  I think zsh has always been implemented like that.  We don't
> really want to change it; the standard (IEEE Std 1003.1, 2004
> Edition) says specifically:

...this has been fixed in their CVS.

> I checked on my Fedora system (about Core 4, I think) and the only
> init
> script without a path is network:
> 
>   cd /etc/sysconfig/network-scripts
> 
>   . network-functions
> 
> If it's still there it probably ought to be fixed.

FWIW, I had tons of those in my FC5's ifup-* scripts.

> It would be relatively easy to interpret these in the same fashion as
> some other form of quoting.  However, there's not much
> internationalization in these parts and won't be without a huge
> effort
> and it's not clear it's worth the effort doing anything apart from
> that.

I agree, supporting those $"funny variables" is probably not a huge
win.

Thanks!



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com