3.4.4pre3 is now availabe from the usual download site. It cleans up the
modperl2 support a bit, fixes a slew of other small bugs and comes with
some other cleanup.
I expect to turn this into 3.4.4 in relatively short order. I know that
there are some additional wrinkles in the query builder that folks have
been finding over the past few days and would love to see patches for
those for RT 3.4.5, which I’d expect in a few weeks unless anything
horrible is found in 3.4.4, but the localization issues, Mason 1.3x
cleanup and mod_perl2 support make a speedy 3.4.4 release pretty
important.
Jesse
Are there plans to roll out a 3.5.3 any time soon?
Yes, but since it’s a dev series and not a production release, it’s a
much lower priority than everything else. If you’re running 3.5 at this
point, you may want to be tracking subversion
Jesse Vincent wrote:
I know that
there are some additional wrinkles in the query builder that folks have
been finding over the past few days and would love to see patches for
those for RT 3.4.5
The one I confirmed contained a patch against 3.4.4rc2 (I believe I had
a typo in my mail). Are there any other reports that need patches?
Rolf
Jesse Vincent wrote:
I know that
there are some additional wrinkles in the query builder that folks have
been finding over the past few days and would love to see patches for
those for RT 3.4.5
The one I confirmed contained a patch against 3.4.4rc2 (I believe I had
a typo in my mail).
I’d kind of wondered. You’d said it was against 3.4.2rc2, which I could
believe. Interestingly, I can’t make RT crash like the original
poster, though I do get weird results trying to add and delete at the
same time.
Are there any other reports that need patches?
Someone posted a very strange AND/OR case just yesterday.
Jesse Vincent wrote:
Jesse Vincent wrote:
I know that
there are some additional wrinkles in the query builder that folks have
been finding over the past few days and would love to see patches for
those for RT 3.4.5
The one I confirmed contained a patch against 3.4.4rc2 (I believe I had
a typo in my mail).
I’d kind of wondered. You’d said it was against 3.4.2rc2, which I could
believe. Interestingly, I can’t make RT crash like the original
poster, though I do get weird results trying to add and delete at the
same time.
Try adding 2 new items, while also deleting 2. It doesn’t “crash” if
you’re only adding or deleteing one.
Someone posted a very strange AND/OR case just yesterday.
Ok, found it and I can also verify that one. And while I have a simple
fix (attached), that I believe is correct, I’m a little surprised that
parser works as well as it does. 
Also, the comment for it refers to a
similar parser, which may or may not have a similar problem … which I
guess I should explain, too
The problem we’re trying to fix here stems from the fact that when
building the expression tree, the aggregator is only set together with a
match criteria. However, for every level or brackets, a new aggregator
is created. So for bracketed expressions like ((a and b) or (c and d)),
with a, b, c and d being criteria) the or-aggregator is never set. The
fix I’m proposing moves the assignment of the aggregator to when it’s
actually encountered in the expression.
Now, I haven’t checked what the other parser is used for, but it really
looks similar. I’ll have to check, but maybe you can give me a hint?
Also, if there’s a plan to rework the parser (it’s the same in 3.5.x),
please let me know.
Rolf.
rt.patch (625 Bytes)
Ok, found it and I can also verify that one. And while I have a simple
fix (attached), that I believe is correct, I’m a little surprised that
parser works as well as it does. Also, the comment for it refers to a
similar parser, which may or may not have a similar problem … which I
guess I should explain, too
There are two TicketSQL parsers. The “original” (and definitive)
lives in Tickets_Overlay_SQL.pm. The Query Builder uses a slightly
“modified” version of the parser that sets up the data structure for
the dynamic stuff needed by the UI.
Also, if there’s a plan to rework the parser (it’s the same in 3.5.x),
please let me know.
There is a plan to rework, rewrite, and unify the parsers. One of
these days it’ll happen.
-R