Empire and D are in the news (page 2)

On 10/20/2021 11:34 PM, rikki cattermole wrote: > The latest bug was with floating point, multiply one number with another *should* result in another and it even writeln's to that value! But it wasn't that value. Which of course meant... the resulting multiplication was like E-6. Tiny but it threw off everything with chromatic adaption. I had a lot of trouble with rounding errors. This was before the intertoobs, and I didn't know there were standard techniques for dealing with it.

On Thursday, 21 October 2021 at 06:34:09 UTC, rikki cattermole wrote: > > [snip] > > Ugh, I've been working on fixing bugs/improving my matrix struct recently. With the bug fixes for inversions/Gaussian elimination. > You can't use LAPACK?

On 21/10/2021 11:01 PM, jmh530 wrote: > On Thursday, 21 October 2021 at 06:34:09 UTC, rikki cattermole wrote: >> [snip] >> >> Ugh, I've been working on fixing bugs/improving my matrix struct recently. With the bug fixes for inversions/Gaussian elimination. >> > > You can't use LAPACK? Nah, need something small and in D. Don't want to complicate things. Don't want to bother with libraries either due to -betterC requirements ext.

October 22, 2021

Re: Empire and D are in the news

Posted by Basile B.
in reply to max haughton

Permalink

Basile B.

Posted in reply to max haughton

Permalink

On Wednesday, 20 October 2021 at 22:47:50 UTC, max haughton wrote:

On Wednesday, 20 October 2021 at 20:33:50 UTC, Walter Bright wrote:

On 10/20/2021 9:13 AM, Basile B. wrote:

On Wednesday, 20 October 2021 at 13:01:21 UTC, Walter Bright wrote:

"Eventually, my clandestine activities were discovered and the computer division demanded that I be reprimanded"

very cool anecdote.

Having upper management go to bat for me against bureaucratic fiefdoms was very gratifying.

I mentioned this anecdote to a friend of mine who is aiming to work in aerospace, he didn't actually believe me. Totally unimaginable today with all these finite element solvers available that run really fast on commodity hardware, although I think the art/intuition of sitting down and actually doing the maths on some mechanical component might be going the way of the dodo.

it's not about the domain, it's more about when this happened, e.g the 80's
It's totally an Halt and Catch Fire story.

the guy has gained access to the mainframe to perform his computations... it's basically too good.

Forums