The race to replace Redis [LWN.net]

The race to replace Redis

Posted Mar 29, 2024 7:06 UTC (Fri) by pbonzini (subscriber, #60935)
In reply to: The race to replace Redis by immibis
Parent article: The race to replace Redis

The difference is that the SSPL breaks the "License Must Not Restrict Other Software" clause of the open source definition. In contrast the GPL and LGPL have an explicit text saying that they don't extend to other software even if distributed on the same medium, if that constitutes "mere aggregation").

(In addition it's debatable whether it's possible at all to distribute all the required components under the SSPL, which would constitute a restriction on making use of the program in a specific field of endeavor; but that's not necessary given the previous point).

to post comments

The race to replace Redis

Posted Apr 1, 2024 16:34 UTC (Mon) by immibis (subscriber, #105511) [Link]

The SSPL also doesn't extend to other software existing on the same medium. I think the best way to view it is that it extends the definition of "linking" for the age of the web app, to close loopholes in the GPL that web apps exploit. It stops allowing you to manipulate "at arm's length" to avoid "linking" two programs that are, in reality, still intimately related.

Is AWS DynamoDB (for example) a work? Common sense says that it's still *a* work, no matter how many executable files it's divided up into. And that work is derived from all of the parts that it's derived from. And if any of those are AGPL, so is the whole thing. Except it's not, because the FSF seems to believe each executable file is a separate work just because the interfaces are documented.

SSPL is supposed to fix that.

It also has some wording problems that may make it unusable, but if that's our only objection then we should make a new license like SSPL but with those problems fixed, instead of rejecting it outright.