RFC on dropping sqlsrv support

markh · March 12, 2021, 7:43pm

Just posted this on GitHub, but would also like to direct attention of integrators and former MAB members to this request for comment on dropping sqlsrv from MODX (possibly 3.0):

github.com/modxcms/revolution

RFC: Dropping support for SQL Server in 3.0 [accepted]

opened 07:40PM - 12 Mar 21 UTC

Mark-H

proposal needs-docs

## Request for comment ### Background For as long as I can remember, Revol…ution has supported an `sqlsrv` driver which allows it to be installed natively on windows servers using a Microsoft SQL Server rather than MySQL. In that same time span, I've probably encountered like... 4 installations actually _using_ MODX on `sqlsrv`, the latest being [this thread on the forum last week](https://community.modx.com/t/extras-not-displaying-on-microsoft-server-install/3778) where a user was trying to figure out why extras wouldn't show up, which is because those extras do not support (or do not claim to support) the `sqlsrv` driver. ### Proposal Given... - There does not appear to be much, if any, real world usage of the sqlsrv driver - Anecdotally I don't believe many (if any at all) core integrators and contributors actively test changes against sqlsrv, at most blindly copying schema changes as needed assuming things work - Unit tests for each commit and pull request (in github actions) are only executed in a MySQL setup - For MODX to be useful on sqlsrv, common extras also need to support for sqlsrv, which requires explicitly providing those schemas and models which I've very rarely seen in the real world (although extras that don't have a custom table are easier to have compatible). ... I'd like to invite feedback on dropping sqlsrv support entirely from MODX as of a future major release. That future major release could very well be 3.0 if there's a clear consensus ahead of beta. That would still give enough time to cleanup the core, update the documentation, etc. ### My opinion Properly supporting sqlsrv would require a much more intensive effort from the entire community (integrators, contributors, **and** extras developers) to consistently test and making sure MODX and extras operate as they should on sqlsrv. I do not believe that effort is worth it given the lack of users, so we might as well just stop pretending it's equally supported as mysql and drop sqlsrv support as of 3.0.

alroniks · March 13, 2021, 10:21pm

I would support such proposal.

alroniks · March 14, 2021, 11:21am

Do not create a flame in the ticket, I also have a question, how we can manage such kind of question with polls? It was a formal process in the MAB, but maybe we can pool active integrators and contributors (simplified MAB), who can make such technical decisions? It would also work during discussions of the roadmap and plans. I am glad to move my post to a different topic if it does make sense.

markh · March 14, 2021, 2:18pm

If a major project like Symfony can make major decisions through a simple GitHub issue with people commenting do we need to make things harder? I’ve tried pushing hard for setting up MAB processes and that clearly didn’t work.

alroniks · March 14, 2021, 3:03pm

Yep, it is why I am suggesting simplify the process, but still have the ability to somehow count and evaluate results. And have an agreement, of course. If we all agree, that we will have a topic, created as an Issue, and people may vote by specific emoji, let’s do it but such away.

I don’t want to have strict and formal rules, but some rules still should be written down just for clarity.