Re: --with-store-dir and/or --localstatedir seem to be ignored

[Taylan Ulrich Bayırlı/Kammer]

Mark H Weaver <address@hidden> writes:

>>             ;; There is no /usr/bin or /bin - replace it with /gnu/store
>>             (substitute* "testsuite/cpio.tests"
>> -              (("/usr/bin") "/gnu/store")
>> -              (("usr") "gnu"))
>> +              (("/usr/bin") (%store-directory))
>> +              (("usr") (car (filter (negate string-null?)
>> +                                    (string-split (%store-directory) 
>> #\/)))))
>
> What is the rationale for replacing "usr" with "gnu" here?  In the
> general case where (%store-directory) might be almost anything, I fail
> to see why "usr" should be replaced with the first component of
> (%store-directory).

The test suite contains:

testing "cpio -p with absolute paths" \
"echo /usr/bin | cpio -dp cpio.testdir 2>&1; echo \$?;
ls cpio.testdir" \
"\
1 blocks
0
usr
" "" ""

Where the second argument is code to evaluate and the third argument the
expected output; it expects the cpio command to print "1 blocks", the
'echo $?' command to print "0", and the ls command to print "usr"
because that directory has been copied there (without its contents).

I believe this works in all cases except where (%store-directory) is /.
Should we try to prevent such a limitation?

Taylan