Allowed maximum package size

[Andrew15-5]

In the forum topic it has been discovered that while some packages were rejected for big file/package size, other packages with even bigger files got merged without a single question about the matter.

As such, there is no definite answer or rule on what is accepted in terms of size, not for the package itself, nor for the individual files in the package. It's not clear if merging bigger packages was an oversight or some internal decision that is still not documented or otherwise publicly stated.

[sihooleebd]

Agreed. Even if it's not a hard-coded limit, having a publicly stated 'recommended maximum' would save everyone a lot of guesswork during the PR process.

[elegaanz]

Yeah, we agree that accepting auto-jrubby despite it having a file that large was a bit of an oversight on our side. On the other hand, this package provides functionality that is arguably much more useful than pyrunner: the latter would have been fairly limited in any case and is not necessary to typeset some documents (in the sense that there are workarounds if you want to embed the output of a Python script in a document, whereas automatic ruby can't be achieved easily other than with a package), which also contributed to us rejecting it and accepting auto-jrubby.

In any case, we want to define an upper limit on total package size that would be enforced from now on. We haven't decided on a specific size yet, but it will be implemented. Once we start enforcing it, the limit will be documented. We may still accept packages that were previously accepted even if they are over the limit such as auto-jrubby to make it possible for it to be updated and for documents using it to continue using the latest packages and Typst versions (but we will encourage its author to investigate how they can reduce the WASM size and do it as much as possible).