[9fans] Bgetrune while(i>runesize) loop

[9fans] Bgetrune while(i>runesize) loop

[Pietro Gagliardi]

Hello. I was looking at the code for the Bgetrune function trying to
implement a readrune() around read(). It works, but without a
counterpart for this:

		while(i > bp->runesize) {
			Bungetc(bp);
			i--;
		}

What is the point of that loop? Thanks.

Re: [9fans] Bgetrune while(i>runesize) loop

[Russ Cox]

> Hello. I was looking at the code for the Bgetrune function trying to
> implement a readrune() around read(). It works, but without a
> counterpart for this:
>
> 		while(i > bp->runesize) {
> 			Bungetc(bp);
> 			i--;
> 		}
>
> What is the point of that loop? Thanks.

Use "\xE0\x00\x00" as your input and you will find out.

Russ

Re: [9fans] Bgetrune while(i>runesize) loop

[Pietro Gagliardi]

I got nothing spectacular from that, but changing the \x00s to QRS
gave me something interesting: without the loop, readrune gives me
[?]S, but with it it, it gives me [?]QRS, like cat would've.

Would this be because fullrune doesn't check for validity, and if
chartorune fails then the possibility of eating too much comes about?

On Jul 10, 2008, at 2:53 PM, Russ Cox wrote:

>> Hello. I was looking at the code for the Bgetrune function trying to
>> implement a readrune() around read(). It works, but without a
>> counterpart for this:
>>
>> 		while(i > bp->runesize) {
>> 			Bungetc(bp);
>> 			i--;
>> 		}
>>
>> What is the point of that loop? Thanks.
>
> Use "\xE0\x00\x00" as your input and you will find out.
>
> Russ
>
>

Re: [9fans] Bgetrune while(i>runesize) loop

[Russ Cox]

> Would this be because fullrune doesn't check for validity, and if
> chartorune fails then the possibility of eating too much comes about?

Mostly, yes.  The fundamental issue is that, presented
with malformed UTF-8, chartorune needs to look at
more bytes than the number it returns.

Russ