A public forum for discussing the design of software, from the user interface to the code architecture. Now closed.
I don't have an exact quote. I think someone like Turing said this, in ye olden days before there was what we now know as programming. It went something like:
The task of instructing [a programmable machine] could never become tedious, because given a tedious task, there would automatically be the interesting task of instructing the machine in the correct way to perform the tedious task for you.
Or something along those lines. Ring any bells?
Nope. But the assumption is wrong. Making instrustion to tell something else how to make instructions isn't inherently more interesting.
son of parnas
Thursday, December 15, 2005
Yeah, but instructions on how to create instructions for creating instructions are /fascinating/. What we need is a general-purpose tool-building factory factory factory.
The tedious part is not in the act of creating code to perform tedious routines. It is in the support stuff that you need to do to make sure it works _properly_ and _well_.
The heart of Barbary Monitor took me almost a trivial amount of time to design and code. Adding the UI, making sure it meets the Windows guidelines, writing unit tests, performing unit tests, debugging, documenting, and so on took a long time and were mostly quite tedious.
Friday, December 16, 2005
Related to this topic.
After I put requirements lists of what kind of compiler I was going to use I realized that with my expertice it would be far better Idea to make my own instead of searching and learning and testing thousands of different ones.
1st requirement truly crossplatform. Nothing more than recompilation for a cross platform operations. There is plenty of frameworks that claim to be crossplatform but truly aren't. There is plenty of minor annoyances that need to be tested and corrected, and with that I'd rather fix my OWN code than large hairy ball of others code.
Java isn't what I need. There is compability problems all over it.
2nd I need to be in charge of optimization, performance is utmost importance.
3rd all those optimizations need to be automated, so that I don't have to 100% of time rewrite those optimizations.
4th it needs to be carbage collected and higher level than C.
Now the other reasons.
I have plenty of knowledge about certain type of compiler writing. And low level operations on target platforms.
I can afford a full year before any income and I can survive with tiny income after that.
I plan to spend 10 years or more improving my program.
If I grow large then there is more possible division of work when part of the application and optimization is done on compiler, and while most is done in the language compiled.
I had my doubts of doing it my self for a while. Then I realized that my plans are not for next 3 months its next 10 years. The solution I decided is not for everyone. Actually most people are better off using ready made platforms. There are clearly defined problems that doing compiler my self can solve. And the fact that I've done them before hand, makes the trade off for me different than for most people starting micro isv.
This topic is archived. No further replies will be accepted.Other recent topics
Powered by FogBugz