Re: gnuclient screwage?
On Thu, 08 Mar 2001 14:57:42 +0900, "Stephen J. Turnbull" said:
>>>>> "Ben" == Ben Wing <ben(a)666.com>
writes:
Ben> i think the proper behavior is to always use /tmp, provided
Ben> it exists and is world-writable; only fall back on
Ben> environment variables otherwise. the reason for this is
Ben> precisely to avoid problems like this -- we want a
Ben> "well-defined location".
How about an app-specific variable like GNUSERV_SOCKETDIR? This would
have higher precedence than /tmp. Then fail if neither is writable.
Either of these solutions would be fine with me, as long as the code and
documentation agree on the behavior. ;)
But yes, for right now, the most pragmatic fix is to just document the
current use of $TMPDIR interpolation.
/Valdis
Attachments:
- attachment.sig (application/pgp-signature — 211 bytes)