Hey, I'm looking to connect to SQL Server, and I see that Stitch/Singer already has a tap for this but it's on the JVM (https://github.com/singer-io/tap-mssql) Does that work with Meltano or is it a deal-breaker? I don't have JVM installed at the moment or I'd test it, but will set it up if it'll work

@sam_woolerton As you've found, Meltano currently only supports pip packages as plugins. @aaronsteers filed an issue a month ago to bring support for tap-mssql through support for Docker images: https://gitlab.com/meltano/meltano/-/issues/2205, but no progress has been made on implementation yet. If you're able to get the tap-mssql executable running on your machine, though, you can trick Meltano into using it by setting your custom

tap-mssql

plugin definition's

executable

property to an absolute path (starting with a

/

) pointing at the executable. Your

pip_url

could point at any arbitrary Pip package, since it will need to be installable, but won't actually be used, because the

executable

property is used when the plugin is actually invoked. Of course, this makes your Meltano project less portable than it otherwise would be, since it'll depend on tap-mssql already being present on the system. Would that be an option for you? Let me know if it works, or if you'd like some additional pointers 🙂

What about this tap for SQL server? https://github.com/wintersrd/pipelinewise-tap-mssql

@cody_kaiser Good find! That should work out of the box with Meltano 🙂

@sam_woolerton Which tap did you end up using for SQL Server? I found 2 available and was going to start using one. Did you try both? https://github.com/singer-io/tap-mssql https://github.com/wintersrd/pipelinewise-tap-mssql

Great find on the Pipelinewise tap, I didn't come across that one. I only tried the JVM one and couldn't get it to work (haven't used JVM before so was a bit out of my depth)

Thanks for update. I'll try the pipelinewise version and see if I have any luck. Right now im using Snowflake streams to capture cdc info after its been loaded to Snowflake. I might stick with that model and look into a tap Snowflake.