Re: Internet protocol question

[Rick Macdonald]

I don't know about any official protocol, but I would say the lock
messages should only be seen when the client does specific lock and unlock
commands:

LKDB
210 GNATS database is now locked.
UNDB
210 GNATS database is now unlocked.
<submit PR>
211 Ok. 
<the report text>
.
210 PR added.

IMHO, when submitting a new PR the lock/unlock activity is something
internal to the submit function and in this case the extra messages should
not appear. Locking is encapsulated in the submit function and the client
has no business or interest in knowing about it or dealing with it. It's
an implementation detail of the submit function. 

Put another way, it may be that a user has authority to call the submit
function from a client but does not have the authority to call LKDB/UNDB
directly. Again, why send lock/unlk messages?

Of course, gnatsd is responsible for ensuring that the lock is released
whether the submit succeeds or fails.

On 4 Sep 2001, Milan Zamazal wrote:

I'm not familiar with Internet protocols, so I need an advice on a
certain gnatsd output.

When a new problem is submitted via gnatsd, its output looks like this:

  subm 
  211 Ok.
  <the report text>
  .
  210 GNATS database is now locked.
  210 PR added.
  210 GNATS database is now unlocked.

or, in the case of failure, like this:

  subm
  211 Ok.
  .
  210 GNATS database is now locked.
  402 Failure reading header
  210 GNATS database is now unlocked.

As you can see above, after submitting the report text gnatsd outputs
*three* status report lines as single messages, the overall result
message is wrapped by the locking status messages.  Is this behavior
and/or format right?  If not, how should the output look?

Thanks for any advice.

Milan Zamazal

...RickM...