Re: redundant DOC files

[Stefan Monnier]

>> Actually, you won't see garbage.  There's a low probability that you'll
>> see the wrong docstring because the mismatch was not detected, and a high
>> probability that the mismatch will be detected and then the docstring
>> pointers are recomputed and things work correctly from then on.
>> The low probability is of the order "1/number of bytes in a docstring",
>> so less than 1%.
> I don't have statistics (as don't you),

This probability is not based on stats but on an analytical model based on
the idea that the docstring pointer into the DOC file is basically a random
offset.  Of course, this is not a truthful model, but it shouldn't be too
far off, if you ignore special cases such as the point to the
first docstring.

> but my experience is that wrong doc strings are seen much more frequently
> than 1%.

Odd.  I've been using a single DOC file often updated by a compilation while
a current Emacs is running, and I've never noticed a wrong docstring, so my
experience suggests much less than 1%, which is admittedly strangely far
from my analytical model.

Anyway, it's not like it matters.  After all, I have no intention to suggest
that we should install the wrong DOC file ;-)


        Stefan