> > Course, it would help if there was a locate command on Solaris,
>
> Dear Sir,
> if you need to use the `locate` command to find
> something, you have a MUCH bigger problem.
Dude
I know more ways to find stuff than I can count. I'm only recanting
the whines from folks who claim that more than one directory in the
path leads to new users not being to find a binary without "locate".
> You shouldn't have any software "installed" on a
> system without the said software coming on the system
> as a package,
Oh, you mean I shouldn't compile any of my own code on my own
system cause I might have to rm -rf it instead of builiding a package
every time. I'm really sure the new users we are catering too
are just *gonna* love that. Course, packaging for Solaris is
quite a bit of magic for the uninitiated. I think I built my first package
in 1994.
> and packages are *easily* queriable down to every
> single file much faster than with `locate` hack.
And the new users we are trying to cater to know this *easily* queriable
command down to the single file, instead of relying on a well known (and
possibly hacky) command called locate. I'm sure if you had considered
the new users we are catering too, you would have included it in your
response, but I think you're just trying to be contradictory.
for the newbies out there, try "grep '<file> f' /var/sadm/install/contents"
(however, this has a typically grep'ism such that any "file" that matches,
will also get all cases for this). This looks for an entry in
/var/sadm/install/contents
where "file" is a "file" (as opposed to a directory or link)
Dude, I think you are missing the point. If you're going to make
Solaris more user friendly to linux users, you don't need just the basic
tools, you need the extras.
> > and some effective HTML documentation detailling out all the tools
> > that exist on the Solaris install (since every argument I've
> > heard about doing this relies on the issue that users can't
> > find the tools)
>
> Again, man pages are much faster to search through
> and locate the correct information, they are also a
> standard that works across all UNIXes, and should be
> used accordingly.
I don't know about you, but the man pages on my current Nevada system suck.
Even when I do a "for i in ${MANPATH}; do; catman -w -M ${i}; done. man -k
typically responds with every possible combination, which was ok 15 years ago,
but now it's completely *worthless* unless you get a "complete match".
> HTML documentation has no place on a UNIX system.
Ah, still got a wyse 50, I see.
> It's hopelessly inefficient compared to UNIX man pages.
That's what I'd call the current man pages. At least with HTML,
pages are easily cross-referenced without having to reinvoke "man",
not to mention forward and backwards are easily understood concepts.
> I know Linux man pages are basically useless babble,
> but one shouldn't assume the same is true on UNIX.
Oh, the "UNIX" man pages. let's see. my /usr/share/man
has approximately 13,000 man pages. Wanna bet that
some of them came from Linux? Wanna bet that some of
them are out of date, have bugs or provide the wrong
information?
> > All this is going to do is further confuse the issue
> > on Solaris WRT buildling open source software.
>
> I believe that was his goal to begin with.
> Unfortunately, in this case the old
>
> "any publicity is good publicity"
>
> will not have the desired effect. On the contrary,
> and one can clearly see here that it's already
> starting to backfire.
Backfire? I think I finally understand the /usr/gnu proposal,
and can see that my thoughts on the topic appear to be
very simlilar to it's constructs.
You want a gnu'ish system, put /usr/gnu/bin before /usr/bin.
You want a Solaris-gnu system, just make sure /usr/bin is before /usr/gnu/bin
> > However, all this will really serve is to make some
> > open source software build a little easier out of the
> > box, but will continue to let folks just assume that
> > the software builds correctly.
>
> Agreed. We are going about it in a completely wrong
> way, trying to react to the effect instead of
> reacting to the cause!
That's not quite what I meant. Lots of folks hack on open
source code from the wild to get it working on Solaris.
Some work in the system, and modify the autoconf/configure,
makefiles and other files in a portable way that allows the
changes to move back upstream. Others just package it
up anyway they can, with no visibility to how it was done
(CCD, SFW, Blastwave). Pkgsrc is about the only one that
is visible, but that's cause it has to built. But most often,
folks just fix it so it works for *them*, and just go on, letting
the knowledge of how they fixed it inside their community
(possibly of one, themselves) as opposed to sharing it with
the Solaris community or the Source themselves.
> The correct way to fix this whole situation is for
> Linux developers to migrate to Solaris, and forget
> about Linux. That would fix all these compilation
> issues.
Dude, You have got to be joking. That is a lofty goal,
and desirable, but it is not going to happen.
Even with a full Gnu-ish set of binaries, Open source just doesn't
build out of the box nearly on Solaris/Open Solaris as well as it
does on other *nix platforms. Solaris is full of idiosyncracies like
/usr/openwin, /usr/dt, /usr/sfw, no Gnu linker, dozens of places
that include files hang out, not to mention the constant extra
package locations for additional features that an Open Source
package might require. I can tell you from years of experience
that unless someone Solaris saavy has been actively porting the
source and propogating the changes, that the configuration of
the source and building of it, is going to require some time
investment between minor (20-30 minutes if you know Solaris
and it's idiosyncrancies) to hours or days. Course, you can
cheat and just overload LD_OPTIONS, LD_LIBRARY_PATH, and
bunch of other stuff just to brute force it, but then it just didn't
compile out of the box.
So you think that just because we have all the tools, that we can
get the Linux folks to come running and drop their Linux boxes
and adopt Solaris. I like the idea. Lofty. Nvidia saw the light
when they realized that they had so much less work to do every
month without having to worry about kernel interfaces changing
under them. But until a majority of open source code just compiles
right out of the <pkg>.tar.{Z,gz,bz2}, we're all going to have to
work to make sure changes get put back upstream to make the
new users coming to Solaris less afraid of buildling their favorite
tool on Solaris.
This message posted from opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
[email protected]
