Thu 09 Oct 2008 06:20:47 PM UTC, original submission:
Harald Bergmann notes that in one place we say "the shell is responsible for wildcard expansion" where later we say "[if you use .o in a] command, wildcard expansion will* take place at that time" (my emphasis on will). Well, perhaps it will, perhaps it will not - that depends on the shell, as noted earlier. Further, any expansion doesn't take place at the time make sees the command, which the above wording could reasonably be interpreted to suggest, but only when make runs the shell and the shell runs the command.
On the mailing list, I suggested [
I'm not sure I've understood. Perhaps rewording the second stanza like
this would address your concern?
"However, if you use the value $(objects) in a target or prerequisite, wildcard expansion will take place there. If you use the value $(objects) in a command, the shell may perform wildcard expansion when the command runs."
] and the OP replied (privately) [
Hi Martin,
I am sure you did!
Your proposal is OK!
Best regards,
Harald
].
That proposal is not quite the same as the patch, from CVS, that I'm attaching here, because the wording has seemingly moved on (using the word "recipe" instead of "command") and because there's quoting in the original that didn't survive the email, leading me to suggest my own $() "quoting", above, which I abandon here in favor of the existing style.
I reproduce it as evidence that my first paragraph here is an accurate summary of the OP's gripe.
|