[DGD] DGD 1.5.3 and Hydra 1.3

[bart at wotf.org]

On Sat, 20 Aug 2016 22:02:53 +0200, Felix A. Croes wrote
> bart at wotf.org wrote:

<snip> 

> > At any rate, it seems kf_read_file (in src/kfun/file.cpp) is using ALLOCA, and
> > hence may crash dgd when configured with a larger string size and trying to
> > read a large file.
> >
> > A solution similar to kf_restore_object seems like a good idea to me.
> 
> I have followed your suggestion, but it's time to start weaning yourself
> away from strings longer than 64K.  Take a look at /lib/String in the
> cloud library to see in which direction I think you should be moving.

I've been looking at such a solution, and it seems desirable, if only to
ensure parse_string can actually generate a parse tree without overrunning the
max array size. 64k may already be too large for that in some cases.

But it also seems to me error strings could easily get longer then 64k when
using them to pass information out of an atomic function (the atomic_message
mechanism) in case of errors. Trying to encode extra information in error
messages is how I ended up wanting a larger max string size to begin with many
years ago.

Bart.
--
http://www.flickr.com/photos/mrobjective/
http://www.om-d.org/