[GRASS-dev] Re: new v.in.geonames: problems with UTF-8 Unicode text
Ivan Shmakov
ivan at theory.asu.ru
Thu Jul 10 13:28:33 EDT 2008
>>>>> Glynn Clements <glynn at gclements.plus.com> writes:
[...]
>>> Yep. This isn't a problem for tables which are created through the
>>> DBMI, as we can ensure that it accepts whatever it creates. The
>>> problem arises if the user creates or modifies tables externally.
>> Then it becomes the user's responsibility not to exploit this
>> SQLite-specific feature (or, otherwise, be ready to solve any
>> inconsistencies that may arise.)
> It isn't just about "SQLite-specific" issues.
Indeed.
> The question is whether sqlite/describe.c only needs to understand
> the types which sqlite/create_table.c itself uses, or whether it
> needs to understand all of the "standard" types (i.e. anything
> defined by one of the SQL standards, as well as any additional types
> which are supported by "popular" RDBMSes).
On the one hand, it certainly makes sense to try to obtain the
most complete description from the source database when
replicating data. On the other hand, it isn't completely
necessary and may even be harmful (consider, e. g., an
application which allocates static N-character string storage
for reading VARCHAR (N) data from SQLite.)
Therefore, it may make sense to provide both and let the
developer decide.
More information about the grass-dev
mailing list