[ Some of you have already seen this; posting to TCLCORE now for
   broader comments. ]

Based on comments seen in the chat, email, and other Tcl development
circles, I've seen folks express interest in tackling a number of issues
related to the way Tcl and its packages get installed from the base
source distribution.  It seems to me the issues are interrelated, and
there may be dependencies that force a particular order in addressing
them.  It also seems to me that the changes could be fairly disruptive,
or at least require significant testing to keep their disruption under
control, so it's better to get things moving in earlier alpha releases
of Tcl 8.6 as opposed to dumping it all into 8.6b1.

A sketch of the issues follow.  These are vague descriptions.  The
precise specs and number and order of TIPs to achieve the whole are
what I hope will arise out of this discussion.

   * Revising Tcl's install conventions so that multiple minor releases
     of Tcl really can coexist on a system ("the tcl.h / tclConfig.sh
     overwrite problem")

   * Revising Tcl's expectations to find packages under something like
     /usr/lib to look (first? instead?) under something like /usr/lib/tcl .
     ("the slow startup problem")

   * Solving the problem that on some platforms the location to install
     arch-independent packages and arch-dependent packages is the same,
     so that then other architectures working from the same filesystem
     either can't get to the arch-independent stuff, or incorrectly get
     arch-dependent stuff for an incompatible arch.  ("The lib64 problem"
     or perhaps "the /usr/share problem")

   * Offering a configuration option (default?) where the parts of Tcl
     that currently must be found at runtime -- init.tcl, encodings,
     message catalogs -- get embedded in libtcl itself as an internal
     VFS.  ("The basekit configuration")

   * Bundling with the Tcl source distribution a larger collection of
     extension source code distributions.  ("batteries included")

   * Completing the TEA-taming of Tk so that it might even be one of
     those extensions in a "batteries included" Tcl source distro.

I'm sorry that at this time I'm not offering a grand road map to get
all this done.  It just seems to me that if all these balls are really
in the air for Tcl 8.6, we need to start catching them, and if it matters
which we catch first, we need to figure that out.  Also we need to
figure out if there are other balls in the same interrelated set that
I haven't listed yet.

Conversely, if it's clear when looking at the complete list that we
just won't be able to get all these done while still getting 8.6 done
by October, then it's good to figure out now what portions, if any,
are within our reach and/or still worth doing in the absence of the
rest.

Several of these matters already have partial solutions implemented by
some of the re-distributors of Tcl, and I think we should steal their
solutions as much as it makes sense to do so.

-- 
| Don Porter          Mathematical and Computational Sciences Division |
| donald.porter@[...].gov             Information Technology Laboratory |
| http://math.nist.gov/~DPorter/                                  NIST |
|______________________________________________________________________|

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Tcl-Core mailing list
Tcl-Core@[...].net
https://lists.sourceforge.net/lists/listinfo/tcl-core