Danek,

Thanks for taking a look.  Comments inline:

Danek Duvall wrote:

> > http://cr.opensolaris.org/~talley/6667461_6667462_6667464/
>
> Everything else in the gate patches up man pages with a sed script.
> Is there a reason you didn't do it the same way?

Yes.  I needed to be able to pass a stability parameter to the script,
which varies depending on the man page.  The sed scripts in other
directories assume a single stability level for all man pages.

> I'll make the same comment as I did for Nico -- ident lines are
> comments in all files that aren't run through a processor which
> understand "#ident" as a pragma.  Thus shell scripts and makefiles
> shouldn't have "#ident", but "# ident", just like any other comment.

You're right, though this nit appears to be pandemic among the
Makefile.sfw files in the SFW gate.

> fetchmail/Makefile.sfw:
>
>   o line 36: This isn't used.  (Same line in the other makefiles.)

PKG is used as a parameter to the sunman.sh script on line 68.

>   o line 44: What's needed out of /usr/sfw/lib that doesn't come
>     from this consolidation (and hence the proto-area version should
>     be used)? (Same question for the other makefiles.)

libssl.so is the dependency that fetchmail needs from /usr/sfw/lib.

I have removed this from the Makefiles that don't require it.

>   o line 88: why are you constructing fetchmailconf in the makefile?
>     Why not just create the file and install it?  Or, since you're
>     doing nothing but execing the python file, why not just put the
>     python file in /usr/bin?  Is that to prevent the .pyc file from
>     getting created in /usr/bin?

This is to overcome a bug in fetchmail's build configuration; the
fetchmailconf created in the build points to python files in a static
location other than the one that I've configured.  Since it is a
one-line script, it is easier to overwrite it than patch it.

> mutt/Makefile.sfw:
>
>   o line 128: I'm not terribly happy with this, but I don't know
>     whether our gettext supports locale aliasing.  Anyone know?
>     Should we file an RFE if it doesn't?
>
>   o line 129: This should probably be /etc/mutt.

I wavered a bit on this, but ultimately chose /usr/share/mutt/etc in
part to avoid conflicts with the system's /etc/mime.types.  But I can
be persuaded here.

>   o line 131: Do you really need this on top of line 130?

Unfortunately, yes.  It is a peculiarity (bug) of mutt's configure
process.  Some parts of the build require --with-docdir, others
--docdir.

> mutt/stubman.sh:
>
>   o line 34: I don't think this manpage needs to be run through eqn.
>     And not tbl as it stands, though it should have the standard
>     table in an ATTRIBUTES section with the Interface Stability
>     entry.

The ATTRIBUTES section with stability is added by the sunman.sh
script, along with the other man pages.

> slang/Makefile.sfw:
>
>   o line 82: likely sysconfdir should be under /etc here, too.

Given that slsh is not going to be widely used, I don't feel it merits
cluttering up /etc with slsh.rc.  As with mutt, I can be persuaded.

>   o line 115: likely not useful for the 64-bit target.

Removed.

>   o line 132: I think you could just do "make install-elf", no?

install-elf also installs usr/bin/slsh (and doc files as well).

> SUNWfetchmail/prototype_com:
>
>   o line 55: We don't typically ship .pyo files.

Why ship the .pyc but not the .pyo?  It seems only the .py is
necessary.

>   Do make sure that the .pyc file has a newer timestamp than the .py
>   file; otherwise it'll get re-created at runtime, and pkgchk will
>   fail.

A quick test of removing both the .pyo and .pyc files, then running
fetchmailconf, shows that neither the .pyo nor .pyc files is
recreated.

> SUNWmutt/protoype_com:
>
>   o line 120ff: Is there a reason that message files are going in
>     /usr/share, rather than /usr/lib?

mutt elects to put them there, and there appears to be precedent:
SUNWpostgr-libs and SUNWgnome-dtlogin-integration, among others, also
prefer /usr/share.

> SUNWprocmail/depend:
>
>   o Is there a reason you're not using the common depend file here?

These are the the direct dependencies I determined that procmail
needs.  If there is a common dependency subset that every package
should use, shouldn't this comment also apply to
SUNW{slang,mutt,fetchmail} as well?  And is this subset documented
somewhere?

> All of these components are at least partly GPLv2.  Typically legal
> requires that we put a disclaimer at the top of the copyright file
> (or something similar, but that's how we've been interpreting that
> requirement) that says that we're distributing under v2 *only*, and
> no later versions.  Were you not asked to do something like this?

Thanks for the pointer -- I did not notice this requirement in the TOI
pdf.  I'll add the preambles to each.  Incidentally, I notice there
are several GPL-licenced packages in SFW that don't carry this
preamble.

Steve
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 185 bytes
Desc: not available
URL: 
<http://mail.opensolaris.org/pipermail/sfwnv-discuss/attachments/20080227/1b16e118/attachment.bin>

Reply via email to