Re: Bindat can exhaust memory when unpacking to vector

[Petteri Hintsanen <petterih@iki.fi>]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> Kind of an embarrassing blind spot.  BinDat should instead encourage the
> addition of sanity checks within the parsing code (including maybe add
> specific support for it).

This was also something I had in mind, along the lines of what Tomas
wrote below:

<tomas@tuxteam.de> writes:

> OTOH, a maximum length of something might be thought as a (n optional)
> part of a type description. Types are predicates.
>
> As could be a set of bounds for a numerical type. Perhaps there's an
> extension opportunity for bindat.

So yes, I also agree that this could be an extension opportunity, but I
wouldn't push it further until we have concrete use cases.  It is
usually futile to try to generalize from zero examples.  (Well, I do
have one example, but still.)

For the time being using inline checks within the specification suffices
for me.

Thanks,
Petteri