Nat!'s Web Journal

2006 has been a very slow year for this Blog, even worse than 2005. There are a few reasons. The most likely is of course Blog-rot. You start writing a Blog, you run out of ideas. The frequency of entries starts to dwindle, readers start losing interest. The Blog folds.

Obviously, that is very similiar to what is happening here...

Some reasons that affect the dearth of submissions.

1. old age is taking its toll, always something to consider. :)
2. my "other" hidden blog is taking up some of the technical submissions. For MulleEOF I have another "secret" blog, where I put my technical problems and solutions. The entries are really only interesting in the context of MulleEOF, so I don't put any effort into writing them, since I am the only one reading 'em.
3. I am mostly coding for MulleEOF, which contrary to its name is proprietary. It's fun to dick around and expand and refine a framework, that basically beats everything else available :) On the plus side, there is the occasional update of MulleEOInterface and MulleSybaseEOAdaptor.
4. the remaining part of free time is taken up, trying to learn guitar, which is pretty time intensive. I might start to write some stuff about guitar playing here. Unfortunately my playing sounds like shit, so don't expect any sound samples here.
5. I don't have any spare time for other little projects, that may end up as Mulle article or product in some form.

I am one of those, that really didn't like the move to the i386 architecture. This blog entry is typed on my trusty G5 machine and I expect this to be my main "coding" machine for some time. The main reason being, that I read PPC assembler much more fluently than i386 and I would therefore lose an advantage, changing the platform.

I fully expect that Apple and third party developers (Adobe) will drive a stake through the heart of the PPC rather sooner then later. Some software will not "work as well" or at all on PPC. It's just the way it goes.

Anyway, I got a few mails regarding apparent incompatibilities of MulleCipher with i386. The SHA-1 is apparently faulty, and ciphered data is not cross-platform compatible. I didn't have access to a i386 machine until fairly recently, and there wasn't much I can do about it. Interestingly, although apparently quite a few people use MulleCipher, and it being open source, I didn't get a fix sent back. This is not unexpected, but the unexpected did not happen, which basically confirms by belief, that open source is only mutual benficial in a very indirect way. And that way is, that it's more a social convenience, I open my code, so you can copy/paste stuff from it, and in turn I benefit from other people opening up their source. If you aren't Firefox or Linux, chances are slim, that you will get back useful code submissions. Anyway, last month I got loaned an 20 Inch Imac and I could reproduce the problems. So far so good. I think I fixed the SHA-1 bug, but the problem with the enciphered text hasn't been solved correctly. If I have a spare day, I will do it, but I have no immediate need for MulleCipher, so it could take a while.

The MulleSybaseEOAdaptor is not compatible with Intel. The core TDS/CT library, the MulleSybase adaptor is based on, is written in C. When I wrote it I didn't think about endianness issues. So a straight recompile won't work. Interestlingly I learned from the Sybase mailing list, that Sybase is "waiting for business case" to port their code to MacIntel. Well this business case, won't very likely appear out of thin air, so the MulleSybaseEOAdaptor could also be used as an alternative C-interface library to Sybase database on Mac OS X... hmm hmm... :) Well there is always FreeTDS, which I haven't touched the last few years. Maybe it has come around.