>>>>> Ivan Shmakov writes:
>>>>> Roger Binns writes:
[…]
>> Consequently if you had a trigger pulling a stunt like this, your
>> code could try to insert a blob and silently (wrongly) end up with a
>> string. SQLite won't even complain if the blob isn't a valid text
>> encoding producing an invalid string.
> I don't have any triggers (at least, it wasn't my intent to add
> them.) My code is roughly as shown below.
> And I don't seem to understand where's the problem.
Silly mistake on my part, as I've just found.
Namely, I've had /two/ functions to alter the table in question.
One of them INSERT's the tuple, and uses sqlite3_bind_blob ().
The other, that UPDATE's the tuple, however, uses
sqlite3_bind_text ().
Before I saw that there's sound support for blobs in the current
SQLite, I've planned to use Base64 for these fields; after, I've
changed one of the functions, but not the other.
And yes, code duplication is clearly a bad thing.
Now that the discrepancy is fixed, the problem is gone.
Thanks.
[…]
--
FSF associate member #7257 Coming soon: Software Freedom Day
http://mail.sf-day.org/lists/listinfo/ planning-ru (ru), sfd-discuss (en)
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
