Rewrite of IBM doublebyte charsets (original) (raw)

Ulf Zibis Ulf.Zibis at gmx.de
Tue May 12 22:45:56 UTC 2009

Am 12.05.2009 21:03, Xueming Shen schrieb:

Ulf Zibis wrote:

*** Question: Why you code: } else if ((byte1 & MSB) == 0) { // ASCII G0 instead of: } else if (byte1 >= 0) { // ASCII G0 I believe this line was written 10 years ago, so I have no idea (or forgot) why we picked this one, my guess is the code might be a little easier to read with "MSB"...you think the >=0 is better or faster/ Yes, I think it's also faster, as loading of "MSB" + AND would be saved. Not sure if HotSpot will detect the shortcut! it actually should be "byte1 < 0x80"

Oops, you are right. I was in the assumption, that byte1 was, "naked" from byte array, in range 0xffffff80..0x7f. Maybe it's better to do: byte byte1 = sa[sp]; or mask by 0xff after the if .. else

... so we could save some more loop effective opcode by "byte1 >= 0"

-Ulf

More information about the core-libs-dev mailing list