Re: unreliability

>> What in particular lead you to believe from your perusal of the code that
>> such problems existed ?
> I see, for example, several uses of dbm_write() to record critical
> information.

What "critical" information ?

The uses I know of for writable DBMs involve hints -- nothing critical.
When things have been bust (e.g. network down, F/S AWOL, etc) I tend to just
"rm" the hint DBMs.

> DBM databases can be corrupted in all sorts of weird and wonderful ways if
> the system crashes.

Are you recommending doing that at each reboot ?

>>> What are you going to say to a user after exim destroys an important piece
>>> of mail that it accepted responsibility for delivering?
>> What leads you to believe I'll need to ?
> The exim design and implementation.

In particular ?

> There is no reliable way to determine where the truncated message ends
> and the next message begins.

It appears to do a pretty goo job of it !
On what you you base your statement ?

If it finds 4 ^As and a newline, it expects either EOF or another 4 ^As and a
newline to indicate the start of the next message.
If it does not, it interprets the first 4 ^As and newline as the *start* of
a new message rather than the end of the previous, i.e. the previous message
is corrupt.

What is the unreliability there ?

> However, the unreliability of mboxes is not what I'm worried about here.

OK -- what *IS* ?

(Boy, it's like trying to get blood out of a stone !
Shall we take this off the list until I've actually been able to extract
some info of interest to the rest of the list ?
)