Tom Hughes

Is that just the first time it is called or is that still true the second time?

Hmmm... Of course if there was actually any decent documentation for the apache module API then it might be easier to figure these things out...

Well the production servers are using mapnik 2.2 so if you're using 2.3 or 3.x then that might explain the differences?

Well that won't alter the XML will it - the XML is input to mapnik. Only the carto version will alter that. Currently both machines are running [email protected] which I...

I should add that removing the layer name from the project file changed the error, revealing what layers mapnik thought were available: ``` 1) #Tile() "before all" hook: Error: OGR...

Looks like that is breaking with the older mapnik that travis uses so I guess this is a breaking change in mapnik...

That'll be it. I was testing in Fedora 28/29 both of which have GDAL 2.2 now.

Well what else is node-mapnik going to pull in? I'm not using any of dane's prebuilt binaries here - it's built from source against the system mapnik. In fact specifically...

The problem as you say is that `spaten` includes WKB data which can be of either endianness, with the first byte being 0 for LE and 1 for BE and...

Oops. So that's a regression that came from a fix for an ARM code generation bug I reported with mapnik ;-) See https://bugzilla.redhat.com/show_bug.cgi?id=1422456 and https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79671 for the somewhat horrifying/incomprehensible history...