Follow-up Comment #14, sr #110901 (group libtool): [comment #13 comment #13:] >> This sounds fine to me. Could you write a patch for this, or specify the ENV >> variable and parts of path conversion that you would like bypassed? > > > I updated > https://github.com/mitchcapper/libtool/compare/add_bypass_path_conversion_opt Thank you for the updated patch, but this will not have the desired behaviour. Additional options are currently being passed to func_convert_core_msys_to_w32 for cygpath, so if LIBTOOL_MSYS_NO_PATH_CONVERSION=1, func_convert_core_msys_to_w32_result would equal "-w" instead of the path.
> and have attached the patch too. In short setting
> LIBTOOL_MSYS_NO_PATH_CONVERSION=1 completely bypasses all conversion.
> Obviously if there is a better name by all means.
I like the idea of adding a path conversion bypass configuration option to
libtool more than relying on an ENV variable. It could be beneficial for other
windows path conversions if it is designed like aix-soname. While it is
annoying to have to modify a source package before building, I think making a
small edit to configure.ac for LT_INIT is okay.
> One additional option for even general improvement could be to simply check
> for the presence of a slash before running the conversion.
I would like to have a check cached for "cmd //c" or "cmd /c", but I am not
sure about how best to test this. It may only lead to libtool hanging much
sooner...
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/support/?110901>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc
Description: PGP signature
