base build fails on PPC Tiger, images_to_archives.tcl barfs

[yaseppochi (Stephen J. Turnbull)]

"base" is at subversion r78893 (don't recall the previous version built, sorry, but it was late last year). iBook G4 (PPC), Mac OS X 10.4.11 (Tiger)

Like so, leaving no error.log I can find:

/usr/bin/tclsh src/images_to_archives.tcl "/opt/local/share/macports/Tcl"
This could take a while...
shell command failed (see log for details)
    while executing
"system "cd $location && $tarcmd -cjf $newlocation * > ${targetdir}/error.log 2>&1""
    ("foreach" body line 72)
    invoked from within
"foreach installed $ilist {
    set iname [lindex $installed 0]
    set iversion [lindex $installed 1]"

[jmroot (Joshua Root)]

Look in /opt/local/var/macports/software/*/error.log

[yaseppochi (Stephen J. Turnbull)]

Of course, as I told you, there is no such file anywhere, as reported by find / -name error.log (which took literally hours).

The problem is that somehow I have a bunch of ports with no image directory and the archive file already built, so the command

"system "cd $location && $tarcmd -cjf $newlocation * > ${targetdir}/error.log 2>&1""

barfs on the "cd $location" command before the error log is created.

OTOH, I also have a bunch of ports with an image directory, so now port -u barfs trying to uninstall them. On top of that, I have a bunch of ports with non-bz2 archives, so the short-circuit of detecting the archive and not trying to build one in that case doesn't work.

[yaseppochi (Stephen J. Turnbull)]

The attached patch works for me (or seems to, I'm up to gcc44 which apparently actually has an image directory that needs to be turned into an archive which will take a while). I'm not a Tcl programmer, various scaffolding has not been cleaned up, no attempt was made to handle the cases the original script tried to handle when I have modified that code. IOW, if you are not prepared to have your brain spontaneously combust, don't look at this code'''

That said, enjoy!

[jmroot (Joshua Root)]

From what I can tell the problems you describe should have been fixed by r79445.

[yaseppochi (Stephen J. Turnbull)]

Replying to jmr@…:

From what I can tell the problems you describe should have been fixed by r79445.

Of course, I updated to r79548 before messing with any code.

Possibly r79445 will prevent it from happening to anyone else in the future, but it sure did not fix things for me.