FreeTDS Development Group

[Frediano Ziglio <freddyz77 AT tin.it>]

Il dom, 2002-09-08 alle 16:33, Brian Bruns ha scritto:
> Now that jkl is putting the final touches on the 0.60 release (I make a 
> terrible release manager as evidenced by the frequency of releases over 
> the years, so I'd like to express my gratitude to Jim for taking over the 
> responsibility) I think it makes sense to start thinking about what 
> features we want in 0.61 and beyond.  If everyone put in their 
> plans/wishes we might be able to plan the upcoming releases better.
> 
> For myself I have two big items and some smaller ones.
> 
> 1) Get a solid corpus of unittests for ODBC, and have it become a top 
> tier API.
> 
> 2) Multibyte support.  There's been some discussion on this, I'd like to 
> see at least UTF-8 supported decently, big5 and ucs-2, etc... might be 
> 0.62 fodder, we'll see I guess.
> 
> 3) output params for dblib are ready as soon as we open 0.61.
> 
> Not specifically 0.61 but futures:
> 
> 1) placeholder support for TDS 5.0.  This is partially done but crashes, I 
> hope to revisit it soon.
> 
> 2) Get connection pooling working with all protocol versions.
> 
> 3) Finish off the TDS dissector for ethereal (while not directly FreeTDS 
> related it'd be a big help).
> 
> So, that's my list, how about yours?
> 
> Brian
> 


- ODBC. I started viewing FreeTDS mainly for this reason. Not so much
time for improve ODBC (I worked mainly on core lib)... Is the worst
library of the group. Some change I'd like to see:
  - improvement of all API
  - many regression test
  - support for dynamic query
  - best error report
  - be able to ignore freetds.conf using SQLDriverConnect, more
compatiblity with windows one (same connection string is useful)
- core tds. 
  - support for multibyte character. Already started on ML but not all
agree with a single implementation...
  - dynamic query, even for mssql
  - cursor support (0.61? perhaps not...)
  - more solid. Test all malloc and report correctly all out-of-memory
problem. It should no core for these problems... Not so difficult but
quite long...
  - limit the copy of data, now we do
    1- copy from wire to tds buffer
    2- copy to tds buffer to temp buffer
    3- copy to temp buffer to client
    I'd like to remove 2 on 0.61 (as 0.53 do). Be able to copy from wire
to client is perhaps possible... but not in 0.61...

freddy77