Am 06/11/15 um 18:20 schrieb Andreas Ringlstetter:
I was referring to the change in analysis.r where the conditionalRemoving the stop was somewhat correct, but a lot of call sites
are now
lacking defensive checks. Would all the sites calling this
function still working properly? It's not just the widgets,
but also every location accessing conf$boundaries.
And is it really a good idea, trying to fix this with a
default parameter bandaid?
For now (as no one wants to add the defensive checks), I
would actually
suggest not changing the get.cycles at all for now.
Instead just add a new predicate, checking whether there are
any ranges
at all, and make the call to get.cycles dependent as well,
still assuming non-empty result set. The predicate is pretty
sure going to be
useful in other locations as well.
hm, are you suggesting something like the following?
if (there_are_cycles()) { return(get.cycles()) } else {
what.to.do.here()? }
execution was introduced by this patch. It's already quite unique in
that it expects get.cycles() to return an empty result set (and not
to fail - silently or verbose).
if (there.are.cycles()) { cycles = get.cycles(); optional = ... # Do
all the optional stuff } else { optional = none; }
For the generic case (no empty result expected, so the behavior of
get.cycles() may be undefined):
assert.true(there.are.cycles()); cycles = get.cycles();
Or just the first line standalone, when accessing some function
deriving from get.cycles() internally.
Same predicate, easy to reuse, doubles as documentation of
prerequisites.
Whats the advantage of storing this data in files and not in thethe data we store in the FS are costly to compute, and it helps
with development if we can re-use the results during debugging
sessions without having to recompute after each small change.
With the exception of a few saves bytes in the file system, what
are the advantages of removing/no creating these files?
database? Respectively, what was the motivation for that, except for
"It was a standalone component once"?
I fully agree that it is expensive to compute, so it should be
stored. But what I'm wondering, is why it didn't go into the
database.
Similar for the forest generated by snatm, even though that is
already improving with the patch from
which means less data is discarded.[codeface] [PATCH 3/3] Add global email analysis to database