Re: [sqlalchemy] Problem with PG ARRAY's ANY() and SA 1.4.36 (maybe related resolution of #6515 ?)

2022-05-01 Thread Mike Bayer
a fix for that is coming through at https://github.com/sqlalchemy/sqlalchemy/issues/7979 though it's best to give types to bindparam() objects if they don't have a value On Sun, May 1, 2022, at 11:53 AM, Mike Bayer wrote: > it looks like the bindparam() when not given a type is being coerced

Re: [sqlalchemy] Problem with PG ARRAY's ANY() and SA 1.4.36 (maybe related resolution of #6515 ?)

2022-05-01 Thread Mike Bayer
it looks like the bindparam() when not given a type is being coerced to thinking it's an ARRAY, which might be a separate issue, but in any case, give it a type so that there is no ambiguity (do you really need to use bindparam() directly?) conn.execute( table.select().where(

Re: [sqlalchemy] Problem with PG ARRAY's ANY() and SA 1.4.36 (maybe related resolution of #6515 ?)

2022-05-01 Thread Mike Bayer
If you could post some code examples on a github discussion that I can run, that would be helpful. the change basically fixed the any/all operations to correctly interpret incoming arguments so that proper casts could be provided. I dont use these operators myself so it could be that

[sqlalchemy] Problem with PG ARRAY's ANY() and SA 1.4.36 (maybe related resolution of #6515 ?)

2022-04-29 Thread Lele Gaifax
Hi all, I'm hitting what is either a regression or a "deprecated" usage of PG ARRAY's any() method, trying to upgrade an app of mine from SA 1.4.35 to SA 1.4.36. The issue is within a quite complex query, with a filter expression like the following: query = query.filter(((CC.languages ==