[PATCH] compiler warning: pointer type in OCTET_STRING_noalloc()

[Neels Hofmeyr]

On Fri, Apr 15, 2016 at 09:49:55PM +0200, Harald Welte wrote:
> Hi Neels,
> 
> On Thu, Apr 14, 2016 at 02:45:56PM +0200, Neels Hofmeyr wrote:
> > It's useless to declare the str arg as const when all the function does is
> > assign it to a non-const pointer.
> 
> The point of doing it this way is that we don't have to write our code
> in a sub-standard way, just because asn1c is written in a way that you
> cannot have 'const' buffers as input into encoding.
> 
> So 'const' here is used as a hint to the programmer, that a read-only
> string is passed into this function, and hence it is input data.
>
> So this was very intentional.  Now you can argue that my line of
> thinking above is basically ignoring the reality of asn1c.  But then,
> don't we have actual real issues to address, than to debate about thing
> slike this?  Sorry...

I would prefer not to debate about this as extensively, but you guys won't stop
bickering ;)

To resolve, I would appreciate if you either apply my patch or add the cast
within the function.

I still disagree and think it is the caller's duty to be aware that the asn1c
struct entry is not const, and that passing such a buffer around possibly
results in overwriting the data. That ugly cast should be in plain sight
wherever it is necessary, not hidden away in a function call, invisible until
your code breaks. I still see only the single caller that employs const data,
and this caller can easily justify casting to non-const, as the use is well
contained.

Feel free to overrule, just please get rid of the warning...

~Neels

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.osmocom.org/pipermail/openbsc/attachments/20160418/650b4f63/attachment.bin>