[75] in Kerberos_V5_Development
Yet more problems with libcc.spec
bjaspan@ATHENA.MIT.EDU (bjaspan@ATHENA.MIT.EDU)
Wed Jan 17 20:37:27 1990
It's amazing how many problems you find in specs once you start
implementing.
The specs for resolve and generate_new say that the cache returned is
not opened. However, this gives us a race condition, as generate_new
can be called multiple times before initialize is called.
The only full solution I can think of offhand is to change the spec so
that the specific interaction between generate_new/resolve and
initialize can be fudged a little bit. In this case, the file for the
cache could actually be created at gen_new/resolve time (even though
nothing would be put in it) thus preventing the race condition.
