On 3/8/07, Chris Kuklewicz wrote:

What happens in your throw/catch case if I have stm1 = do some_stm_code_that_throws_your_exception stm2 = return Foo and I run "atomically (stm1 `orElse` stm2)" ? Answer: The exception will prevent running stm2.

In the *specific* case of the external dictionary code, there is no orElse case to worry about since the algorithms are deterministic in this sense. For a more general case, it is an interesting problem to consider. You might want something like throwIfNothingElseSeemsLikeAGoodIdea, so that you can attempt any alternatives, and if none succeed then throw (and I guess choose the first if more than one alternative wants to throw - I like determinism).

...

Um, is unsafeIOToSTM $ atomically trans going to run you into problems?

YES!

My point is that onRetry really wants to be not "IO t" but "IO_execpt_STM t", but there's no way of ensuring that, which unfortunately antagonizes modularity in the same kind of way that locks do. It requires you to know how the implementation of the IO actions you use work, in case they use STM inside. Since the actions may come out of a library somewhere, you need to know how the library is implemented. <sigh> Why are tricky problems tricky? I thought Haskell was supposed to make everything easy. ;-) T. -- Dr Thomas Conway You are beautiful; but learn to work, drtomc@gmail.com for you cannot eat your beauty. -- Congo proverb