Garrett-R
commented
4 years ago I personally like your choice of requiring including the --as-index flag, but curious to hear @c4software's take on that.
One idea is that if the sitemap ends up being 50K+ links, then at the end of the run, if they ran it without --as-index, you could output a note to the terminal with a link to some educational resource on why they should consider breaking it into multiple sitemaps with --as-index.
c4software
As first mentioned in this issue, sitemaps over 50,000 URLs should be split into multiple sitemap files, with a single master index file pointing to all of the sitemap files. This PR adds support for outputting a sitemap index and multiple sitemap files.
Right now,
--outputis not a required parameter; but outputting an index and multiple sitemap files without writing them to files isn't quite sensible. The index contains pointers to files, so what would be the contents of the index when no output file is specified? Because of this, I'm currently requiring--outputwhen using the new--as-indexflag.In order to output an index, you have to include the
--as-indexflag. If you don't include the--as-indexflag, then the sitemap will be written to a single file, even if there are more than 50,000 URLs. My thinking was this would maintain backward compatibility; presumably everyone using the library right now is happy with the output so why change it? Another possibility would be to take this away as an option and just always write an index if there are more than 50,000 URLs, since this would be in line with the specification. If we do go with this, we would likely need to make --output required due to the first bullet described above.