This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Patch to remove useless non_lvalue calls


>>>>> "Joseph" == Joseph S Myers <jsm@polyomino.org.uk> writes:

Joseph> [...] but once these languages are fixed we can get rid of
Joseph> NON_LVALUE_EXPR, reduce the number of different conversion
Joseph> tree codes about, stop creating the unnecessary
Joseph> NON_LVALUE_EXPR tree nodes and avoid all the checks for this
Joseph> tree code.

Speaking of which, it seems to me that the distinction between
NOP_EXPR and CONVERT_EXPR might be valuable to the optimizers, but it
doesn't really make sense for a front end writer.  It seems to me that
in GENERIC we could have just one of them, and then let some later
pass differentiate as needed.  Am I missing something here?  I've
always assumed this is just a historical oddity, but perhaps there's
some reason for it I haven't figured out.

I'm sure that sooner or later I'll introduce a bug along these lines
into my front end :-)

Tom


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]