This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: validate_replace_rtx_1 fix
> I'm afraid I have more grammar nitpicks for you:
No problem. I will try to improve my english language skills :)
> > + /* If only one operand is a `neg', `not',
> > + `mult', `plus', or `minus' expression, it will be the first
> > + operand. */
>
> It *will be* implies it already is, which is not the case.
This sentense is copied from manual.
Can we say:
/* If only one operand is a `neg', `not',
`mult', `plus', or `minus' expression, make it the first
operand with highest preference. */
>
> /* If only one operand is a NEG, NOT, MULT, PLUS, or MINUS expression,
> put that operand first. */
>
> Maybe you should also say why these get priority over ordinary complex
> expressions (simplifier does better?)
Basically it allows you to simplify insn patterns, where complex expressions
are on the both arms of commutative operation, as you know in what order they
come.
I've run across such example while constructing pattern for double shift.
Honza
>
> > + /* Complex expressions should be the first. */
>
> No "the" in this sentence. /* Complex operands should come first */
> is consistent with the rest of the comments.
>
> > + /* Return 1 iff it is neccesary to swap operands of commutative operation
> > + in order to canonicalize expression. */
>
> Needs an 'a' before 'commutative' and a 'the' before 'canonicalize'.
>
> --
> zw "Ultimate Mortal Kombat 3" ... is feminist in depicting a woman as fully
> capable of holding her own in violent combat with heavily armed men.
> -- American Amusement Machine Association vs. Indianapolis