Gadfly server startup error

[Ralf Muschall]

Steve Holden wrote:

> Yes, the whole Gadfly SF web seems to be a bit of a mess.

Maybe it has been infected by the source code+doc ;-)

I just started the server (after applying Carl's changes)
with

gfserver 2222 foodb . ''

(the DB "foodb" had been created earlier using gfplus
locally).

If I look for it with "nmap localhost -p 2222", it enters
an endless loop (I don't know whether it would still work
after this, since I could not get it to work at all).

gfplus in client-mode wants the argument order
  DB, port-No, password, host
(inferred from the source, i.e. the manual is wrong).

The argument order of commandlines and functions seems to be
a general problem - maybe I should try all 24 possibilities
(almost each place in the source does them differently).

After that, it says

|gfplus $Revision: 1.7 $ -- Interactive gadfly shell  [readline]
|
|Wednesday September 13, 2006 04:53 PM
|Using:
|Policy: foodb
|Loc: localhost:2222

i.e. it includes a thing called "Policy" (unexplained, just
mentioned in the "Security" chapter of the server manual)
and fills it with the value of the database name (but comparing
scripts/gfplus.py and client.py, "database" and "policy" seem
to be synonyms (but this disagrees with server.py)).

Entering a lone semicolon+return at the client prompt causes
a timeout in the client and a message

|no such policy: foodb
|main loop on <socket._socketobject object at 0x4080838c> <gadfly.database.gadfly instance at 0x40435eac>
|server: waiting for connection(s)

in the server (which is true in some sense, since I never made
a policy - I just want a blood DB).

It seems the proper way to use gadfly over the net is to say

echo 'SQL statement;' | ssh -l remoteuser 'gfplus -localoptions'

:-(

Is anybody out there who has used the server+client operation
mode successfully?

Ralf