Kommentare zu: New C++ Database Library in the Make http://blogs.newphoria.de/2008/11/01/new-c-database-library-in-the-make/ Newphoria Project News Sun, 20 May 2012 20:48:29 +0000 http://wordpress.org/?v=2.2.1 Von: ekkehard http://blogs.newphoria.de/2008/11/01/new-c-database-library-in-the-make/#comment-4 ekkehard Sat, 15 Nov 2008 14:34:00 +0000 http://blogs.newphoria.de/2008/11/01/new-c-database-library-in-the-make/#comment-4 BTW, in the meantime, I have used a more traditional solution to templates. The support of "export" has been removed. Instead, the template implementation is included only where needed, and there's exactly one module for every template instance. In all other cases, templates are just declared, not implemented, when #included. This achieves the greatest possible portability. Use of namespaces is optional. Compilers without "stdint.h" and "stdbool.h" are supported. BTW, in the meantime, I have used a more traditional solution to templates. The support of “export” has been removed. Instead, the template implementation is included only where needed, and there’s exactly one module for every template instance. In all other cases, templates are just declared, not implemented, when #included. This achieves the greatest possible portability.

Use of namespaces is optional. Compilers without “stdint.h” and “stdbool.h” are supported.

]]> Von: ekkehard http://blogs.newphoria.de/2008/11/01/new-c-database-library-in-the-make/#comment-3 ekkehard Fri, 14 Nov 2008 15:38:06 +0000 http://blogs.newphoria.de/2008/11/01/new-c-database-library-in-the-make/#comment-3 ADB is making good progress, I have finished a lot of components and the basic (on-disk) memory management is already working. The resulting database files are entirely architecture-independent. :) 64-bit arithmetic is required, however. But this is no problem on any current platform. Data type support for records include 8 to 64 bit integers, signed and unsigned, and multibyte and wide character strings. Floating-point is not inherently supported, because it is not necessary (fixed-point arithmetic does a better job for business applications). There is also the problem that representing floating-point in a platform-independent manner is somewhat difficult, character strings being arguably the best form. Wide characters are only supported as 16-bit and 32-bit representations. ADB is making good progress, I have finished a lot of components and the basic (on-disk) memory management is already working. The resulting database files are entirely architecture-independent. :)

64-bit arithmetic is required, however. But this is no problem on any current platform.

Data type support for records include 8 to 64 bit integers, signed and unsigned, and multibyte and wide character strings. Floating-point is not inherently supported, because it is not necessary (fixed-point arithmetic does a better job for business applications). There is also the problem that representing floating-point in a platform-independent manner is somewhat difficult, character strings being arguably the best form. Wide characters are only supported as 16-bit and 32-bit representations.

]]>