> The rationale after the existence of that function, and also after the
> existence of two different types pdf_fsys_t and pdf_fsys_impl_s, is
> that pdf_fsys_t is intended to be opaque. We may want to expand its
> definition to hold data not related to an specific implementation.
>
Well, then according to the current filesystem module interface,
what abstraction is supposed I should provide in the HTTP filesystem?
a) A pdf_fsys_t variable initialized at compile time and, therefore,
assuming the internal structure of the pdf_fsys_t type.
b) A set of procedures and a pdf_fsys_impl_s structure, delegating
to the user the work of create/destroy the HTTP filesystem type
from them.
The second. The user then can use the create function to instantiate
a filesystem variable.
---
Jose E. Marchesi [email protected]
GNU Project http://www.gnu.org
