On Mon, 2 Nov 1998, Mark Gaulin wrote:
> My point was that is would be great to have the source code open via a remote
> RCS system, so someone could simply check out the files they wanted to change
> and have the rest of the code come out of the "official" source code
You must be joking. First, it's an enormous amout of overhead
that, IMHO, doesn't solve any real problem. Second, it's a
HUGE security hole to allow anybody to check-out, change, and
CHECK BACK IN TO THE OFFICIAL SOURCE unchecked modifications.
What's to stop some person with malicious intent from
introducing bad code?
> I figured C++ is the way to go from here on out [since] that effort had
> substantial benefits such as improved performance and robustness when dealing
> with large indexes.
You clearly don't understand software. There is no correlation
between the choice of language and performance and robustness.
(I mean this for languages in the same "class" such as C and
C++, but I would also put Pascal, Ada, and FORTRAN into the
same class. Obviously, you're not going to get the same
performance out of interpreted languages.)
You can write very fast and robust programs in C. I *could*
have written SWISH++ in C and it would have been as fast and
robust as it is now. I chose to write SWISH++ in C++ because
it was easier for me to do so (especially with the
standardization of STL and my ability to put a wrapper around
> From what I've seen you did a great job with swish++ and moving both efforts
> into one common project could have been of value to everyone.
SWISH++'s source is open for the world to see and use. It uses
a better (O(lg n), worst case, rather than O(n), worst case)
algorithm and mmap for fast I/O. The SWISH-E team can read
that source code and use it if they so choose.
Received on Mon Nov 2 11:42:14 1998