I'm on my second day of porting SQLite to an embedded
environment. I'm not sure if this is the appropriate
place to put these comments about the experience.
I've seen some type of "ticket" mechanism in CVS but I
don't know if these comments raise to the level of a
"bug", etc.
Anyway, three things:
1. There are many instances where (char *) to(unsigned
char *) implied casts are done (and vice versa).
These should really be fixed so fewer compiler
warnings are generated. Perhaps I'm the only one who
likes my builds to have no warnings...
2. There is an implied availability of ctype, stdargs,
stdio, etc. All of these #includes and references to
the routines used within them should be moved to the
OS file. It is not always the case that these
routines are available and in fact, are not available
on my platform.
3. I turned on almost all of the OMIT definitions
since size is very important on my embedded platform.
This causes a few problems with definitions in
parse.c. This is documented somewhere and I expected
it. It shouldn't be necessary to hand-code out the
parse.c code for triggers, etc. though. It would be
much easier if, based on the various OMIT definitions,
stub routines would be defined that force a parse
error, etc.
The porting effort has now turned to a testing
effort...
__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com
