Danek Duvall wrote:
On Fri, Mar 06, 2009 at 05:55:52PM -0800, Brock Pytlik wrote:
Should I include/allow the -a/-p command options which duplicate the
functionality of putting '<>' or not around the entire query? Right now,
I'm leaning towards removing these options and simply relying on <>. The
reason for having -a/-p would be if we wanted to make returning lists of
packages the default behavior, and have users have to specify -a to get
the traditional behavior.
Why do you have the "<>" syntax at all? I would completely expect to
define the result type with a flag to the search subcommand. What happens
if you search for "<token1> AND token2"?
'<tok1> AND tok2' is a syntax error. There's a different semantic betwen
'<tok1> AND <tok2>' and '<tok1 and tok2>' That's why '<>' is needed.
A third option exists for -p which might align better with a users
expectations. -p could mean do pkg conversion on EACH of the highest
level queries I perform. Right now, pkg search -p foo AND bar is
equivalent to pkg search '<foo AND bar>'. What might be better is to
have -p mean '<foo> AND <bar>' (and by extension, do the package
conversion to queries at all levels). Luckily, these are purely client
side issues, so it's something that can be changed easily without
worrying about communication between client and server.
I think you should put a couple of examples of the new search stuff in
pkg(1). In fact, you might have to clean up the search example that's
there; I'm not sure if it's correct now.
Yep, plan to. I thought I'd get the technical stuff out first since
changing documentation carries less risk.
Also, right now I'm packaing ply up as part of SUNWipkg since it's needed
for the client, but not the server. If that's not the right packaging
approach to take, please let me know. I wasn't sure if I should but it into
a second package or not.
It should probably be in its own package. SUNWply is fine for now.
Ok, I'll make that change.
Do you have a deadline for code review comments?
That depends on a number of things I don't have a firm handle on.
Ideally, I'd like to put back monday, but I recognize the chances of us
converging on something that fast is unlikely. I will be checking email
and if I get comments back, I'll respin to try and keep things going along.
Danek
Brock
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
