On Wed, Oct 21, 2015 at 11:43 PM, Mike Bayer <mike...@zzzcomputing.com> wrote:
> If we OTOH use native_unicode=False and use the Unicode type for the > columns we care about, that seems a lot faster, e.g. changing the > mapping to: [...] > When we use the Unicode type and set use_native_unicode=False, > SQLAlchemy's own use of Python unicode codecs takes place; this takes > place within the C extensions and has been highly optimized for speed in > the way that the codec is invoked. We've already observed that > cx_Oracle's native unicode methodologies turn out to be much slower than > SQLAlchemy's, which was unexpected since cx_Oracle is pure C code, and > in this case psycopg2 is also pure C code so it is again surprising > SQLAlchemy's approach is a little faster. Back then when I created the C extensions, I found that out as well: virtually all backends were slower at decoding unicode than our own unicode handling. At that time, there was no easy way (that I knew of, at least) to turn off the native decoding of the DBAPI via SQLAlchemy. I am glad to see that it now exists ! Congratulations to Mike for his dedication and endurance in the project, You never cease to impress. Gaëtan -- You received this message because you are subscribed to the Google Groups "sqlalchemy" group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/d/optout.