Manipulating the package finding algorithm?
- From: Volker Hetzer <firstname.lastname@xxxxxxxx>
- Date: Fri, 19 Jan 2007 13:57:22 +0100
Hi!
We've got a rather difficult software deployment problem.
We are customizing a large PCB layout and schematic program and we need
for a set of tcl programs to load different "incarnations" of packages.
For instance, given a package Xxx, there could be directories
Xxx.BETA.PCB_PROJECT_A.USER_B
Xxx.STABLE
etc., all containing Package Xxx.
And, upon startup, we'd like first to load a bunch of generic packages
like tcom and oratcl and then set up the package discovery algorithm
so that it only checks a bunch of PCB and user specific directories:
package require tcom
package require oratcl
getUsername
getPCBname
doPackagemagic
package require Xxx
where Xxx could be sourced from either of the above directories
depending on PCBname and username.
The alternatives are
- to define my own load procedure but this would force another coding and file
layout convention on programmers and IMHO the package mechanism is a good
one. I wouldn't like to replace it.
- to really change the package /names/ instead of just the directory names.
This would mean fiddling with source code during test and release and I
would like to avoid this. Particularly since the guy releasing from beta to
final isn't necessarily the developer but someone who has no idea about tcl.
Lots of Greetings!
Volker
--
For email replies, please substitute the obvious.
.
- Follow-Ups:
- Re: Manipulating the package finding algorithm?
- From: Georgios Petasis
- Re: Manipulating the package finding algorithm?
- From: Mark Janssen
- Re: Manipulating the package finding algorithm?
- Prev by Date: Tcl_RegisterChannel: duplicate channel names
- Next by Date: Re: Multithreaded DLL in Singlethreaded TCL App
- Previous by thread: Tcl_RegisterChannel: duplicate channel names
- Next by thread: Re: Manipulating the package finding algorithm?
- Index(es):
Relevant Pages
|
