Reset corrupted PStore (fixes #409) #418
Conversation
|
CC @bobthecow |
denisdefreyne
changed the title
| @loaded = true | ||
| rescue | ||
| FileUtils.rm_f(filename) | ||
| load |
There was a problem hiding this comment.
i don't like an this rescue, since we're calling #load again inside the handler. anything that fails which isn't fixable by removing the pstore will just go into an infinite loop. Can we rescue a specific error?
There was a problem hiding this comment.
Good point.
We can rescue TypeError, but to be 100% safe, the #load method should probably take an argument that indicates whether or not this invocation is a retry, and if it is, do not retry but just reraise the error. What do you think?
There was a problem hiding this comment.
Hmm, it’s actually not easy at all to make this fail (I noticed while trying to write a test for this). The only thing where this double-call check is necessary is when FileUtils.rm_f(filename) fails, because in the next call, the function will exit before the begin/rescue/end block, because !File.file?(filename).
So I think it’s OK to stick with the original implementation.
Reset corrupted PStore (fixes #409)
|
Merged anyway, but you can always give more feedback later :) |
This PR makes a store reset itself when corruption is detected. This is fine, because the stores only contain data to speed up site compilation.
Fix for #409.