Hallo Ralf, Thanks for the review.
On 10 Jun 2010, at 23:54, Ralf Wildenhues wrote: > * Gary V. Vaughan wrote on Thu, Jun 10, 2010 at 05:18:08PM CEST: >> Okay to push? > > Some of the $ECHO use ($ECHO "$opt") now became plain echo, that might > be a problem with backslashes or leading single hyphens. Otherwise, it > looks to me like with > --ltdl=foo --ltdl=bar > > the old version used the last, whereas the new one uses the first > option. Agreed, and fixed. > It is a bit confusing to see things like $NL2SP initialized later in the > script than it is used, although that is probably ok since the use is > inside a function. Not lexically later than used, just textually further down the script. > If that's easily fixed though, that might be better? > Or is it that we're guaranteed there is no out-of-function code until > the very end? You decide. For the coupling between general.m4sh/getopt.m4sh and clients to remain loose, the out-of-function code they supply must not be affected by the client out-of-fuction initialization code. My intent with moving it was to have only the code required by the option parse loop (the first out-of-function section in libtoolize.m4sh) read before the loop runs - ostensibly to make --help and the like available as quickly as possible: though in practice a few dozen lines of initialization won't make any difference, on principle it still seems like a good idea. > OK with those things fixed, and if the testsuite still passes. It does. Will push presently. >> * libtoolize.m4sh: Replace hand written shell code with a >> call to M4SH_GETOPTS. Move some premature initialization >> from the preamble to the main part of the script. Exit with >> an error on spurious additional non-option arguments. >> (envopts): Integrate LIBTOOLIZE_OPTIONS pre-parsing into the >> main option parsing loop. >> (opt_copy): Use in place of and in the reverse sense of the >> old opt_link variable. Cheers, -- Gary V. Vaughan (g...@gnu.org)
