From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jeffrey A Law <law@cygnus.com>
To: Mark Klein <mklein@dis.com>
Cc: gcc@gcc.gnu.org
Subject: Re: MPE Port 
Date: Thu, 30 Sep 1999 18:02:00 -0000
Message-ID: <3141.937226376@upchuck.cygnus.com>
References: <4.1.19990908063104.00c24100@garfield.dis.com>
X-SW-Source: 1999-09n/msg00612.html
Message-ID: <19990930180200.yY9FDepmHYoNQWyBoi4Q6kqV52jLbz9xR0ntYWc0mlA@z>

  In message < 4.1.19990908063104.00c24100@garfield.dis.com >you write:
  > Yes, but it suffered from the same problem. (I think MKS is using shared
  > source for both tar and cpio). The problem is that it tries to store/copy
  > . and ..  and that results in a "duplicate" in the destination and ends
  >  with an error. I started to look at porting GNU tar, but there are file
  > characteristics (meta data) that needs to get "attached" to each file,
  > and I wasn't sure how to do that. I've figured out what HP tar does and
  > may be able to extend GNU tar to do the same ... but still haven't tried
  > it.
Ugh.

Presumably there's little chance of getting HP to fix their braindead tools?

If we indeed to have to work around the braindead tar, we may be able to do 
so in a cleaner fashion.

It seems to me the only issue is that tar is crapping out early with an error
and that you could create a new make rule for your braindead tar in your x-*
fragment file, then set up configure.in to use it.

It would model the existing install-headers-tar fragment closely, except that
it would ignore errors from tar.

jeff