Hi.
I just filed https://phabricator.wikimedia.org/T99483 about reconsidering the "pages-articles" XML dumps that we currently generate. I'd be interested in any thoughts or feedback about the current setup and ways to potentially improve it. I suggested one possible approach: splitting by page.page_namespace instead.
If someone could forward this to the XML dumps mailing list and any other mailing lists that seem relevant (wikitext-l?), that would be great.
MZMcBride