I would recommend using pgloader (http://pgloader.io/) instead. It is simple to use in the default case, but can be configured with rules for how data types should be converted if necessary. It can also clean up weird timestamps like 0000-00-00 00:00:00.
No foreign key support is kind of a deal-breaker. Hard to believe anyone who is actually in the position of needing this would have a schema so simple that it contains no foreign keys.
We were getting really close to converting our RDS instance from MySQL -> PG, but now with Amazon Aurora being the spiffy super-scalable/etc. RDS engine, we'll probably end up moving to that when/if needed.
Which is a mixed blessing; Aurora sounds great, but I was looking forward to getting back into PG after years of RDS being MySQL-only.
Yeah, seems like postgres is the only database that doesn't support those things. I mean, I use postgres, but I curse the lack of this feature regularly.
[+] [-] jeltz|11 years ago|reply
[+] [-] adamferguson|11 years ago|reply
[+] [-] scrollaway|11 years ago|reply
[+] [-] anacleto|11 years ago|reply
[+] [-] bjacobel|11 years ago|reply
[+] [-] maligree|11 years ago|reply
or complex
[+] [-] warbiscuit|11 years ago|reply
[+] [-] stephen|11 years ago|reply
Which is a mixed blessing; Aurora sounds great, but I was looking forward to getting back into PG after years of RDS being MySQL-only.
[+] [-] codexon|11 years ago|reply
[+] [-] tiglionabbit|11 years ago|reply
[+] [-] rpedela|11 years ago|reply
[+] [-] caioariede|11 years ago|reply
Works very well for me.
I think the only limitation is related to spatial data. I'm not aware of anything else.
[+] [-] pornel|11 years ago|reply
Fortunately there was a project that reimplemented many of these functions in Postgres:
https://github.com/pornel/mysqlcompat
[+] [-] chx|11 years ago|reply
[+] [-] mihailshumilov|11 years ago|reply
[+] [-] putna|11 years ago|reply
does the job for me
[+] [-] pgib|11 years ago|reply
[+] [-] pmelendez|11 years ago|reply
[+] [-] sarciszewski|11 years ago|reply