TSXxxx functions: die!

Dimitrie O. Paun dimi at cs.toronto.edu
Tue Feb 5 09:58:52 CST 2002

Once again,

Since this list is (again!) so quiet, I feel obligated to stir the spirits
a little. So here it goes: we should get rid of the TSXxxx functions.
    -- we don't keep to standard API, which is against the Wine's spirit
    -- they introduce yet another (useless) layer
    -- we reaquire the lock like crazy for no good reason
    -- most modern X systems are already thread safe
    -- we integrate Wine's locking with glic's, so we should be OK
    -- they are ugly
    rm -rf tsx
    + misc Makefile cleanups
But I hear people screaming: how about portability? What about platforms
where we don't integrate with libc's locking, or when we don't have a
thread safe X? Well, we can create a wrapper driver which simply aquires
the lock, calls into the regular driver, then releases the lock. In fact,
I am not convinced we shouldn't deal this way with the current driver, to
simplify the code. It's not like we're missing a lot on performance...


More information about the wine-devel mailing list