API for science apps

[Nicolas Alvarez]

Do we support compiling science apps against Synecdoche API?

If we do: the last changes to filesys.C break compatibility; high chance
science apps rely on boinc_file_exists(const char*).

(I don't mind if we don't support it, just asking. No strong opinion in
either direction from me.)

[Michael Tughan]

I can try compiling/linking AI's app against Synecdoche in a day or
two to test if we want to.

Michael Tughan
Michael's Programming

[Der Meister]

Nicolas Alvarez wrote:
> If we do: the last changes to filesys.C break compatibility; high chance
> science apps rely on boinc_file_exists(const char*).

Only C-applications should rely on boinc_file_exists(const char*), as
for them the declaration of boinc_file_exists(const std::string&) will
generate a compiler error. However, C++-applications still can pass a
const char* to boinc_file_exists(const std::string&) as an implicit
conversion is possible.

Therefore the question is if we should still support science
applications written in native C. If yes I would suggest to turn the
main interface into a C++-interface and provide a legacy interface for C
based on the C++ interface.