#3 Update on "Further optimization"

[Anonymous]

I tried to post this under "Add a comment below" but it
doesn't look like it worked. If it did, mea culpa for a
duplicate comment. Also, I've added a bit to the end in
a P.S.

Just quickly checking over pngcrush.c (v1.5.8), it looks
like simply changing the "<=" to "<" in line 2554 should
accomplish what I'm suggesting.

Just another thought while I was digging through the
source: Would it be simple to tell whether, even if
chunk alterations were requested, that the output file
actually has not beealtered at all from the input file? For
instance, I often specify -alla, -allb, -reduce, &c on
groups of PNGs, many of which already have only
required chunks & actually have their IDATs at a better
size than PNGCrush determines is best. In these
cases, PNGCrush doesn't copy infile to outfile, but
leaves its best attempt in outfile. Overall, obviously, it
would be better to simply copy the better IDAT chunk
from the infile (if possible), perform whatever other chunk
operations (if any), & drop THAT in outfile, right? Or am
I asking for too much?

Just a few thoughts that you can think through (if
convenient) while working on other stuff, & thanks again,

TSamuel

P.S: Just quickly looking it over again, it seems that
you're trying to handle that last suggestion via
things_have_changed . Is reasonable to assume that
you're working what I said, but haven't quite
implemented it yet?