test please delete

[WTFH]

Afternoon all

Spirited drive to Ferndown to deliver 3 guineafowl.
Sun's out and the temp is creeping up. There was a bit of rain on the journey - unsurprising as I had washed The Toy yesterday evening.

[BR14]

Afternoon all

Spirited drive to Ferndown to deliver 3 guineafowl.
.

long way for someone's lunch

[NickFitz]

Showers happening now

Not cooling down much in here yet though. I've got the fan in the open kitchen window, pointing inwards, in the hope it draws in air from outside. Then I'll move the fan to try to get some of that air from the kitchen moved through the hall in the direction of the living room. Probably a fool's errand, but worth a try

[BrilloPad]

[BR14]

[NickFitz]

[BR14]

[NickFitz]

Just spent ages trying to work out why a set of numbers, passed through a function twice, gave different results the second time. (This was in testing; usually, they won't be the same numbers.)

The pipeline through which these numbers were travelling was from being attributes in an XML file, whence they were read in Python and dumped to a JSON file, whence they were loaded by JavaScript. They define the bounding box of a map in latitude and longitude, aka EPSG:4326, and are reprojected using ProjJS into EPSG:3857. They define the extent of the single layer containing that map.

Later, the minimum and maximum bounds are also reprojected to provide the extent of the layer group containing all the maps; I restricted it to one (of four) maps for testing, so it was the same values. But the reprojected bounding boxes for the layer and the layer group had wildly different minimum and maximum values for Y, which made no sense at all

Eventually I realised that when I serialised them to JSON, they were still strings, as XML attribute values are strings (even if they look like numbers). So the layer bounding box got them as strings, and for some reason I can't be bothered to look into ProjJS gave incorrect results for the Y values, though not the X values. Then, in doing the min/max thing to get the bounding box for the entire group of layers, JavaScript automatically type-converted them to numbers. And when ProjJS got those, it was fine and gave the right answer.

So the bug is in my Python code that doesn't parse them as floats before serialising the data. Alternatively, the bug is in ProjJS, which does type conversion on the X values but not the Y values. Either way, I just edited the JSON file to make them numbers in there and now my stuff works

[BR14]

WTF?

[NickFitz]

WTF?

That's what I said