I wonder if I can do this efficiently in sqlite:

for a row, and one column (most likely blob type),  I want to append
some value to the end of the blob, I do this many times. then I may
read up all the appended sections and write out the entire blob with a
new value

without any special support, I can read existing blob, add some new
data to it, and update the column with the new blob value.
but because most incremental updates are small, reading the majority
of the old blob is a waste of time.

ideally this is a sparse table ( each append creates a new "column",
there is no limit on how many columns each row could have, or any
schema for columns, for that matter). But I can't find an
off-the-shelf solution for sparse tables. (can't use Columnar  DB,
since my access pattern is still first based on rows)

Thanks a lot

sqlite-users mailing list

Reply via email to