Make astyle differ upon astyle version

[colivier]

Differences on styling are present upon astyle versions.

1.22 and 1.23 seems to have same indent behaviour, as 1.24 seems different.

With 1.24 on r5282 return: (same file with 1.22/1.23 is unchanged)

Index: postgis/geography_inout.c
===================================================================
--- postgis/geography_inout.c	(revision 5282)
+++ postgis/geography_inout.c	(working copy)
@@ -138,9 +138,9 @@
 	if ( typmod_type > 0 &&
 	        /* GEOMETRYCOLLECTION column can hold any kind of collection */
 	        ((typmod_type == COLLECTIONTYPE && ! (lwgeom_type == COLLECTIONTYPE ||
-	                                              lwgeom_type == MULTIPOLYGONTYPE ||
-	                                              lwgeom_type == MULTIPOINTTYPE ||
-	                                              lwgeom_type == MULTILINETYPE )) ||
+	                lwgeom_type == MULTIPOLYGONTYPE ||
+	                lwgeom_type == MULTIPOINTTYPE ||
+	                lwgeom_type == MULTILINETYPE )) ||
 	         /* Other types must be strictly equal. */
 	         (typmod_type != lwgeom_type)) )
 	{

[pramsey]

This is even worse for 1.21, Jeff Adams tried to make astyle and it formatted every single file. We might have to remove make astyle and go back to just doing piecemeal formatting and a bulk run for major releases. And perhaps get that commit hook put it.

[jadams]

What about having "make astyle" check for astyle version, and refuse to run if it isn't "whatever version we currently think is best"? Folks can either upgrade, or not run it (which they wouldn't do if you remove it from the makefile anyway).

[strk]

+1

[colivier]

Commited such a test as 5315 with 1.23 as the 'blessed' version

If no one complain, close this ticket by the end of the day