On 2003-07-16T09:31+0100, Max Bowsher wrote: ) OK, I've found the problem. Setup's integrated tar code is rather ) oversensitive to magic numbers. Specifically, your naim tarball does not ) contain "ustar\040\040\0" at offset 257. Repacking with GNU tar is the short ) term fix. Long term, maybe setup could become more tolerant in its tar ) code - PTC. What kind of tar did you pack these packages with?
It was actually packaged on a Cygwin system, using its GNU tar 1.13.25. tbz2: CYGWIN-PATCHES/dist mkdir "${top_builddir}/tmp" make install-strip DESTDIR="`cd ${top_builddir}/tmp; pwd`" (cd "${top_builddir}/tmp"; $(AMTAR) cof - --owner 0 --group 0 usr | bzip2 -9 -c >../${pkgtbz2}) rm -rf "${top_builddir}/tmp" The -o option I pulled from elsewhere in automake's generated files: all automake-generated tar files (including the source tarballs generated by "make dist") will be in that format as well: -o, --old-archive, --portability write a V7 format archive I believe it's done for compatibility with older versions of tar, and I believe that is the format Solaris' tar uses as well. It's ironic that using -o breaks setup.exe's tar routines :) For the next release, should I remove the -o? Also, is there a way to run setup.exe against a local file (not in any setup.ini) to check whether setup can install the package? I'm thinking that could be the last step in the packaging procedure, before uploading to a webserver/submitting to cygwin-apps. Thanks, all, -- Daniel Reed <[EMAIL PROTECTED]> http://shell.n.ml.org/n/ It is so easy to miss pretty trivial solutions to problems deemed complicated. The goal of a scientist is to find an interesting problem, and live off it for a while. The goal of an engineer is to evade interesting problems :) -- Vadim Antonov <[EMAIL PROTECTED]> on NANOG http://site.n.ml.org/download/20030715233159/naim/naim-0.11.5.9.cyg11.tar.gz