From: John V. <ja...@gm...> - 2007-01-23 01:31:09
|
On 1/23/07, Brian Dessent <br...@de...> wrote: > > "Salazar, German P21322" wrote: > > > ...from my (ignorant) point of view, being a simple user of these > > systems (mingw and msys) I feel like if the command "ln" does not > > do in msys what does in a REAL POSIX system (whatever the reason), > > then it should not be made available in the first palce...it just > > lends itself for missleading us users. > > > > So, if there is a command named "ln", I expect it to behave like > > "ln"and not like "cp"...that's all I am saying. > > The whole reason that MSYS exists is to provide an environment in which > to run autoconf configure scripts and execute Makefiles from POSIX-type > software packages with the MinGW toolchain. Many of these expect to be > able to use common POSIX commands like 'ln'. If MSYS did not provide a > 'ln' binary then it would not be able to run some of these configure > tests properly, which would defeat its whole point of existance. By > providing a 'ln' that just copies, it at least allows these configure > tests to succeed in the majority of cases. I agree that ln -s should remain in its current form, but perhaps it should emit a message on stderr to indicate it what it is doing. Also, currently when the -v (verbose) argument is supplied, ln -s on msys will incorrectly inform the user that it is creating a symlink, using the same message that is used on platforms that do support symbolic links. $ ln -s oldfile newfile create symbolic link `newfile' to `oldfile' -- John |