[openssl-dev] [openssl.org #4243] 1.1.0-pre2: bug: EVP_CIPHER_CTX isn't completely opaque

[balducci@units.it via RT]

> This is according to our interpretation of "type opacity", meaning that the
> type name is available but not its content. "Data hiding" is another way to p
ut
> it. This means that there will be a need to adapt, stack allocated
> EVP_CIPHER_CTX is no longer allowed, but there are functions to allocate and
> free them on/from the heap (EVP_CIPHER_CTX_new and EVP_CIPHER_CTX_free).
> 
> What's your interpretation of type opacity? I'm not going to say that you're
> wrong and we're perfect, we're interested in feedback and constructive
> arguments. I just wanted to let you know our thinking so far.
> 

> >
> > apologies if I am missing something here.
> >

...it is evident that I was missing something!

A simple search for "opaque type" brings me to, eg, wikipedia (yes, that
popular!) where I find:

        "In computer science, an opaque data type is a data type
        whose concrete data structure is not defined in an
        interface."

ie exactly what I naively called "inconsistency" (!)

I'm sorry for the noise: I promise next time I'll double check before
disturbing

thanks for your patience and politeness

ciao
gabriele