Sunsetting the XSLT 1.0 DocBook Stylesheets
Published by Occasionally, I observe issues opened on the XSLT 1.0 stylesheets. That makes me uncomfortable, because I don’t think anyone is ever going to fix them.
The first commit in the git repository for the XSLT 1.0 stylesheets occurred more than 25 years ago. The commit message on that commit, “New repository initialized by cvs2svn”, is clearly a migration from CVS to SVN that occurred before the migration from SVN to GIT, but it can’t have occurred very much before. XSLT didn’t exist before 1998.
The last commit in the XSLT 1.0 repository is from 2020. That’s almost 20 years after the first working draft of XSLT 2.0 and is 13 years after the XSLT 2.0 Recommendation. It’s 3 years after the XSLT 3.0 Recommendation. Work is now progressing on XSLT 4.0.
If you’re using the 1.0 stylesheets and they’re getting the job done for you, I think that’s great! I don’t think you should feel obligated to upgrade just because a newer version exists. An old tool that does the job is every bit as good as a new one. (In this age of slopware, maybe better.)
At the same time, I think it’s only fair to set reasonable expectations about what maintenance might be done and whether reporting bugs or providing patches is likely to be useful. In this case, I don’t think it is.
I think it may be time to archive the XSLT 1.0 stylesheets repository and accept that none of the maintainers have plans to continue working on them.
Comments
The problem is not the specs, it's the software. There's one sort-of-open XSLT > 1 implementation for Java which has some impressively hacky bindings for C and Python. Looks like there's also some independent work happening in Rust now. But for most purposes, XSLT after 1.0 is a one-vendor product.
I'm inclined to take issue with some of your characterizations, I'm not sure why you think the Saxon HE release is only "sort of" open and I think the current C, C++, C#, Python, and PHP bindings are all solid.
But I am sympathetic to your position. I wish there were more open implementations of later versions of the standards. Here's hoping some of the current energy leads to fruition.
Unfortunately, no one is interested in continuing to maintain the 1.0 stylesheets, and I think it's only fair to make that clear to potential users.