07.05.2016, 20:10, Kurt Schwehr kirjoitti:
This is cart before the horse but... as fast as I can so expect
typos. Now just think of a ~1K long function or method with tons of
instances and lots of places to bailout successfully or as failures.
We have > 9K free/CPLFree/CPLdelete/CPLDestroys that could be < ~100.
I am definitely for having support for unique/smart pointers to objects.
In my experience one of the things making code difficult to understand
(and thus maintain/enhance) is long subroutines. Smart pointers would
often help a lot in that respect.
Ari
GDALMyBigObj *poInstance = CPLCalloc(sizeof(GDALMyBigObj);
...
if (oops) {
CPLDelete(poInstance)
return;
}
...
return; // kaboom
...
if (oops) {
CPLDelete(poInstance)
return;
}
...
if (oops) {
CPLDelete(poInstance)
return;
}
CPLDelete(poInstance)
return;
or worse
GDALMyBigObj *poInstance = CPLCalloc(sizeof(GDALMyBigObj);
...
if (oops) {
goto END;
}
...
// Careful what you do here because you are crossing gotos.
...
if (oops) {
goto END;
}
...
if (oops) {
goto END;
}
...
END:
CPLDelete(poInstance)
return;
when you could have... And yes, getting to this is not trivial.
There are multiple things here to discuss:
auto poInstance = std::make_unique<GDALMyBigObj>(arg1, arg2, arg3);
if(oops)
return; // Everything cleaned up nice
...
if(oops)
return; // Everything cleaned up nice
...
if(oops)
return; // Everything cleaned up nice
...
return; // Woohoo! Success!
My example of the deleter comes from real code where I don't what a
heap checker barfing at me when ever a test fails. Makes writing and
debugging tests insane....
// Apache 2.0 license...
{
// ...
unique_ptr<OGRFeature> feature(layer->GetNextFeature());
OGRGeometry *geometry = feature->GetGeometryRef();
ASSERT_NE(nullptr, geometry);
char *wkt_tmp = nullptr;
ASSERT_EQ(OGRERR_NONE, geometry->exportToWkt(&wkt_tmp));
std::unique_ptr<char, CplFreeDeleter> wkt(wkt_tmp);
wkt_tmp = nullptr;
ASSERT_STREQ("POINT (-109.27 10.3)", wkt.get());
}
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/gdal-dev
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/gdal-dev
