Call for help testing 1.6.1 revamped postflight script

Juan Manuel Palacios jmpp at macports.org
Sun Jan 13 21:47:59 PST 2008


	As some of you may have seen, I've been improving the postflight  
script in the release_1_6 branch with the feedback I've received so  
far, plus some other relevant fixes/improvements. The final product is  
what will be in the 1.6.1 pkg installer, which I plan to upload to our  
website to replace the buggy 1.6.0 installers.

	I've tested it locally on both virgin and customized accounts,  
extensively, and have found it to work reliably so far. I'd now like  
to openly call for some wider testing in case I'm missing bugs that my  
environment is not surfacing.

	If you're up for it, please grab the script off this URL [1] and take  
it for a hard ride on whatever environment you can think of. I've  
tried to bullet-proof it for straight forward functionality in a  
default environment and to be as polite and non-disruptive as possible  
in a non-default one.

	Standard disclaimer applies about this code potentially having bugs  
that may disrupt your working environment, so do not use on a  
production machine/account if you can't afford any errors.

	For those of you wondering what I'm considering as a default  
environment, have a read at:

http://guide.macports.org/#installing.binary.postflight.details

	That, plus the fact that I only do it for bash and tcsh shells, and  
the latter only as legacy support for Jaguar and previous accounts.  
I'm somewhat considering removing such support because tweaking tcsh  
configuration file has proven a tad difficult:

1) It's not easy to invoke a non-interactive, login tcsh shell  
session. In a nutshell, so far it has proven impossible for me. This  
limits my ability to properly test the environment to figure out what  
I need to add and what I don't, which takes me onto 2) below
2) It's not entirely clear to me whether I should write to ~/.tchsrc,  
~/.cshrc or ~/.login (I've heard solid claims for all of them),  
barring my inability to properly test the environment;
3) barring 2), the form of the settings to be written varies ("set foo  
= bar" Vs. "setenv foo bar");
4) I'm by no means a tcsh user, so all I can do is *guess* the best  
approach in unpleasant tcsh debugging sessions ;-)

	Various instances of 4), plus very kind help from both Eric Hall and  
Wilfredo Sanchez at times, have lead me to a combination of ~/.tcshrc  
with "setenv foo bar" statements, which I think works fairly well. But  
still, I'd love some experienced tcsh using eyes if available to shed  
some more light on this aspect of the script (thinking a bit more  
about it, resolving 1) above would probably solve this entire problem).

	So, without any further ado, I'd appreciate as much help I can get in  
testing this script and, in case of failure, getting detailed reports  
if possible. Success reports are also of course welcomed!

	Regards,...


-jmpp


[1] http://trac.macports.org/projects/macports/browser/branches/release_1_6/base/portmgr/dmg/postflight?format=ra


More information about the macports-dev mailing list