Conversation
|
I think this solution has merits. The other would be to change that definition of |
dot-asm
changed the title
|
WIP is removed, I've verified this with aix-gcc and aix64-gcc... |
|
On related side note, I feel that I'd prefer to harmonize usage of some of the parameters. As discussed in referred issue, there are arflags and ARFLAGS. Then there are ar and AR, ranlib and RANLIB, maybe something else. One case should be sufficient in all these cases. BTW, on side note to side note. My understanding is that one can omit |
I agree, with an exception... for flags, we keep the user overridable ones in upper case attributes and the non-overridable ones in lower case attributes. How that matters for flags to
100% correct. |
It's totally appropriate for cflags and alike, but I suggest that it's an overkill for arflags and alike. In other words I'm talking about harmonization specifically for flags/attributes like arflags, not for all flags. On second thought one might find having for example ranlib vs. RANLIB usable. (Once again, for example). But then it should be possible to actually "wire" ranlib directly into rule. In mean if RANLIB is defined, then rule would invoke $(RANLIB), but if ranlib is defined, then rule should invoke specifically $target{ranlib}, i.e. without referencing the make variable. So I suppose one can also put it as "harmonization to upper case" vs. "hardwiring lower-case into rules". |
Reviewed-by: Rich Salz <[email protected]> Reviewed-by: Richard Levitte <[email protected]> (Merged from openssl#5713)
4 participants
I'm adding WIP, because it's just one way to solve the problem, and it's open question if it's
moremost suitable. See #5659 for background info. On additional not, manual says that ranlib needs -X option too. I don't seem to have the problem, but it might be problematic on older systems...