Re: [Cscope-devel] Bug #449808 - making cscope.in.out endian-safe

[Hans-Bernhard B. <br...@ph...>]

Darlene Wong wrote:

> When  will this problem be fixed?  

"Predictions are always hard, especially if they concern the future."
(A. Einstein).

Seriously, I believe this particular bug is a lost cause.  The code 
dealing with the inverted indices is *horrible*.  I'd much rather dump 
it completely than try to make any non-trivial modifications to it.  I 
actually did so, in a private test version.  I have a copy of cscope 
that uses BerkeleyDB to hold the inverted index.  That's still not 
endian-safe, but it could be made so without too much hassle.

The main reasons I never checked that in are

1) it's a rather massive modification, and I don't really have the time 
or resources to test it

2) it's slower than the current code

3) the BerkeleyDB index is significantly larger than the current one
(factor 2 and more)

Previous experience (from porting the scanner to flex) taught me that 
the cscope user community is strongly allergic to any change that slows 
down the program or makes its databases any larger, so they would almost 
certainly would not have liked this change.