Re: ZSH performance regression in 5.8.1.2-test

[Jun T <takimoto-j@kba.biglobe.ne.jp>]

> 2022/04/27 18:34, Peter Stephenson <p.w.stephenson@ntlworld.com> wrote:
> 
>> On 27 April 2022 at 01:38 Bart Schaefer <schaefer@brasslantern.com> wrote:
>> 
>> I was pretty sure I'd found a case (on Ubuntu 20.04) where it
>> succeeded on a pipe when there was already data written to the pipe
>> before the seek was attempted.  That was only/exactly for (0,
>> SEEK_CUR).
> 
> A simple test for this still caused the seek on the type to fail
> here, but I could be missing the  key elements.  Should still
> be a safe test with only a few bytes in the pipe, I would think.

OK, I will add this to configure.ac.


> 2022/04/26 23:31, I wrote:
> 
> But at least on my Mac the following seems to work also:
> 
>    if (lseek(SHIN, 0, SEEK_CUR) == 0)

Sorry, it should have been:
    if (lseek(SHIN, 0, SEEK_CUR) != (off_t)_1)

I will modify Bart's patch and post it with the patch for configure.ac.