Committed. Commit message probably is too large for automatic approval ;) On 24-10-2006 14:43:23 +0200, Stefan Manegold wrote:
Sounds great to me; hence, I stringly vote for "go" --- ASAP to have enough time before the release to sort-out potential problems/conflicts, etc.
Stefan
On Tue, Oct 24, 2006 at 01:28:34PM +0200, Fabian Groffen wrote:
I have a prepared a change for SQL's meta-data headers to use a different symbol than the widely accepted hash (#) used for comments in MonetDB. Instead, the percentage sign (%) is used to indicate a structured meta-data header.
This change allows JDBC to be strict in its parsing, while it also supports comments/info/warnings over the protocol. Mapi has also been patched to support this.
An additional advantage of this change is that meta-data no longer is considered a comment by Mtest; as such changes in this meta-data (which usually upset or change the behaviour of JDBC) are immediately visible, even though there is no JDBC test available.
In the changes implemented one or multiple comment lines can follow after a regular Response, just before the prompt or next start of header. I changed the signatures of the functions that create the Responses such that they can take an extra string argument "w" (to comply with Niels' coding style) which represents a warning message. This string may contain newlines, in which case the string is properly send as multiple comment lines over the stream.
Before I do this massive checkin (lots of mapproves in SQL) I prefer to get a go/no-go first. In case of a no-go, I will check in my changes on a branch.