[PROJ] The state of PROJ C++ 2020

[Martin Desruisseaux]

Hello Søren

Le 07/05/2020 à 18:24, Søren Holm a écrit :

> I do however not understand why the precision and future evolution readiness -
> which are features in inself - is brough into a purely language oriented
> question. I'm quite sure that using C or C++ without the stuff that makes it
> slow can be just as precise as anything else.

The discussion about precision was an attempt to explain why PROJ 6 
become larger and in some circumstances slower. Improving the precision 
required a more extensive use of EPSG geodetic database than what PROJ 4 
did. Answering to the question "what is the transformation from A to B" 
requires a search in the database. PROJ 4 by contrast was trying to 
answer this question with a little bit of logic in the code, which was 
insufficient (the CSV files were not provided required information). It 
seems hard to me to get the functionality of relatively complex queries 
in a relational database with only a few changes in PROJ 4 C code, 
without SQLite or something similar.

The discussion about future evolution was an attempt to explain why PROJ 
6 got a C++ API much larger than the previous C API. This C++ API is a 
consequence of the decision to follow OGC/ISO standards, which are a bit 
large but usually for good reasons I think. The weight of ISO/OGC 
standards sometime looks like useless complexity, but my experience in 
20 years of use of those standards is that (in my opinion) the 
geodesists who designed them have done a pretty good job of addressing 
"real world" problems that, as developers, we hardly imagine before years.

     Martin


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20200507/1c81fd20/attachment-0001.html>