[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: TEST_ABANDON && TEST_CANCEL

Kurt D. Zeilenga wrote:

At 05:50 AM 4/19/2005, Pierangelo Masarati wrote:


Howard,

I find the capability of testing abandon/cancel by means of ldapsearch
quite useful and very little intrusive; so, as ldapsearch is a "toy"
client, I suggest we enable it by default, possibly by means of a -k
{abandon|cancel} switch. Unfortunately both -k and -K are already in use
by (deprecated) LDAPv2 kerberos options, and the only switch available to
ldapsearch seems to be -r.




I would presume that it would be nice to extend this toy
to other LDAP commands, such as ldapmodify(1).   If so,
then it would be best to use an argument not used by
any LDAP command.

I'm thinking:
       -e abandon
       -e cancel

Or, if use of -e here is a stretch, then:
       -o abandon
       -o cancel

I rather not assign a single letter argument specifically
to abandon/cancel. I rather use an extensible single
letter argument.



I'd object that other tools execute a single operation with a single response, and abandoning/cancelling it sounds a bit useless because it's very unlikely that one can issue the operation quickly enough to get to the point where abandon is checked, if it is at all; ldapsearch also performs a single operation (or a sequence, in case of paged results), but with many intermediate responses. In any case you're right, it could be implemented in any of the commands (if we use the asynchronous call). Actually -e sounds like a stretch, but that wouldn't be too bad; -o is "reserved" (what is it for, by the way?).

p.

--
Pierangelo Masarati
mailto:pierangelo.masarati@sys-net.it



   SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497