Hi Massimo, do you have a link to the SQL injection issue? I couldn't reproduce it, nor the communication problem (there were an out of sync statement issue under high loads, IIRC)
BTW, I was given access to the pg8000 official repository (now it is being maintained again), so I'm planning to merge my version with the latest updates (including some performance enhancements). Joe: I attended the pypy tutorial at PyCon US 2012, seeking to speed up pg8000 without luck. Not only there was no improvement, also I got stuck by a pypy unsuported feature in Windows. Maybe pypy has better support now, maybe the new enhancements in pg8000 are better for its JIT compiler. If you just have to upload a CSV file, see the COPY statement, it is unbeatable. Best regards, Mariano Reingart http://www.sistemasagiles.com.ar http://reingart.blogspot.com On Thu, May 30, 2013 at 6:33 PM, Massimo Di Pierro <[email protected]> wrote: > Mind I have security concern about pg8000. It is vulnerable to SQL > injections in web2py. > > > On Thursday, 30 May 2013 14:41:55 UTC-5, Joe Barnhart wrote: >> >> I have just tried both drivers -- but in an apples-and-oranges comparison. >> I used pg8000 with pypy and web2py because it is pure Python and can be used >> with pypy. I used psycopg2 with python 2.7 on the same database and >> application. >> >> My application begins with a bulk-load of a CSV file. The file has about >> 450,000 records of about 10 fields each. Inserting the file using psycopg2 >> and python 2.7 took about 4-5 minutes on a quad-core i7 iMac. The memory >> used was about 20M for postgres (largest thread) and about an equal amount >> for python. The task was handled by the web2py scheduler. >> >> The pypy-pg8000 version of the file load took almost an hour, but that is >> deceptive. The problem is that it overwhelmed the 12GB of memory in the >> computer. Both the pypy task and the postgres task ran amok with memory >> requirements. The postgres task took >8GB and forced the computer into >> swapping, killing the response time. >> >> Pypy is known for being somewhat of a memory hog (I was trying version >> 2.0.2). It worked darned well in web2py, with this being the only problem I >> encountered. Since my code heavily relies on modules, the speedup was >> noticible using pypy. Some of my longer tasks include creating pdf files >> and this took about 1/3 to 1/5 the time under pypy as compared to cpython >> 2.7.1. >> >> I know this is not an accurate comparison (because of the pypy component), >> but the runaway memory use of postgres under pg8000 concerned me so I >> thought I'd mention it. >> >> -- Joe B. >> >> On Wednesday, May 1, 2013 4:59:26 PM UTC-7, Marco Tulio wrote: >>> >>> Are there any advantages on one or another or are they basically the same >>> thing? >>> I'm using psycopg2 atm. >>> >>> -- >>> []'s >>> Marco Tulio > > -- > > --- > You received this message because you are subscribed to the Google Groups > "web2py-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/groups/opt_out. > > -- --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
