Re: lynx-dev z (was Re: Timeout parameter?)

[Jason F. McBrayer]

>>>>> "BL" == Bela Lubkin <address@hidden> writes:

BL> Are there any readers who know about and deliberately do not turn
BL> on NSL_FORK?

I have it turned on on Linux but off on OS/2.  Under OS/2, neither
fork() nor anything much like it is in the kernel; the fork() in EMX
(unix-like libraries for OS/2) is extremely inefficient.  NSL_FORK
could be efficiently implemented on OS/2 using threads (I did
something similar in porting fetchmail), but I don't have time to
implement it right now.

BL> There *is* an overhead -- the process forks.  Any "modern" Unix
BL> system (using virtual memory as good as 4.2BSD's or better, i.e.
BL> dating back at least 15 years) should take only a small memory
BL> hit...

Is there enough overhead to make NSL_FORK a concern on systems running
a lot of Lynxes at once?  I'm thinking in particular of a Freenet that
uses Lynx as its main user interface.

-- 
+----------------------------------------------------------------+
| Jason F. McBrayer              address@hidden |
| The scalloped tatters of the King in Yellow must hide Yhtill   |   
| forever.                    R.W. Chambers _The King in Yellow_ |