Looking for Linux (Loki) Descent 3 checksum bug fix

[TechPro]

Can't seem to find the checksum bug fix for the Linux Descent 3.

Most of the links in the threads that mention it no longer work except one ... and that download has a problem. I get an error trying to extract the files and cannot get them.

Anybody have it?

[Aus-RED-5]

Sorry Tech, I have no idea what files or links you've tried.

Doesn't help that I don't know much about Linux either. hehe

However, FunkyStickman might be the guy to help.

Also here is a link to files and a how-to guild.
http://jeffhendricks.net/?page_id=63

[Jeff250]

Wasn't the checksum bug caused by a discrepancy between the sqrt() functions on linux and Windows? This page seems to be the best information I've been able to scrounge up on it. If so, wouldn't it be possible to just LD_PRELOAD a sqrt() function that behaves a la the Windows one? If so, that would seem a heck of a lot more elegant than Monk's solution. Unfortunately, information on the specifics of the checksum bug seems to be sparse.

[Duper]

I thought Kevin had that tracked down. Something about the way the math was being done by the engine in the wrong order?

oh wait... you mean compiling, right?

[Grendel]

Ask here.

[akula65]

I have the following:

d3fix-1.6-2.i386.rpm (1.5 MB)
d3fix-1.6.tgz (1.5 MB)

I have been wondering where to get the Perl server script to add new missions to the database (d3fixadd.pl).

[Jeff250]

Wasn't the checksum bug caused by a discrepancy between the sqrt() functions on linux and Windows? This page seems to be the best information I've been able to scrounge up on it. If so, wouldn't it be possible to just LD_PRELOAD a sqrt() function that behaves a la the Windows one? If so, that would seem a heck of a lot more elegant than Monk's solution. Unfortunately, information on the specifics of the checksum bug seems to be sparse.

It looks like x86 C compilers tend to just inline sqrt(), since (1) it's always safe to do so, since it's a standard-defined function, and (2) the x86 instruction set has a built-in sqrt instruction. Which makes me wonder all the more how the sqrt() functions between Windows and Linux could have possibly differed... maybe a single- vs. double-precision thing?

[BUBBALOU]

not that hard

Can't seem to find the checksum bug fix for the Linux Descent 3.

http://descentbb.net/viewtopic.php?t=9855

Unfortunately the topic to the thread that leads you to it is not insightful and the authors links no longer work, therefore it's an informational thread only