On 18 Nov (02:07), dosapati wrote:
> When I use Hibernate session for long duration transactions, Can I
> disconnect the SQL
> connection from the session without committing and reconnect again when I
> need. And finally
> will commit the transaction after all the actions with the system. I will
>
It does not pose much of a problem for the grammar,
especially if the tokens are such that this can be
done with the lexer. Then it is simply a matter of
plugging in the pass through token into the grammar
where it is appropriate (e.g. as an expression
'atom').
Don't know what it means for the ba
--- Chris Nokleberg <[EMAIL PROTECTED]> wrote:
> On Mon, Nov 17, 2003 at 11:54:54AM -0800, Joshua
> Davis wrote:
> I suppose you could do something like
> raw "current date"
>
> So 'raw' becomes a keyword that modifies the next
> token.
Yes. Users of an old system that I worked on that
transla
--- "Les A. Hazlewood" <[EMAIL PROTECTED]> wrote:
> > have a prefix that indicates it should flow
> through into the resulting
> > xml
>
> I think you meant resulting sql, right? ;)
>
Errr... yeah. Did somebody say XML?
__
Do you Yahoo!?
Protect your identity wi
When I use Hibernate session for long duration transactions, Can I
disconnect the SQL
connection from the session without committing and reconnect again when I
need. And finally
will commit the transaction after all the actions with the system. I will
rollback If I get any problems
in the transacti
> have a prefix that indicates it should flow through into the resulting
> xml
I think you meant resulting sql, right? ;)
Les
---
This SF. Net email is sponsored by: GoToMyPC
GoToMyPC is the fast, easy and secure way to access your computer fr
On Mon, Nov 17, 2003 at 11:54:54AM -0800, Joshua Davis wrote:
> That's a neat idea. So, 'raw:' would be a
> pass-through that would pass things directly to the
> underlying engine. The only issue there is when the
> underlying syntax does not obey the rules of the
> over-arching grammar (e.g. 'cu
> I would rather you go the other way, i.e. unprefixed functions should be
> portable ANSI-derived aliases which are mapped by the dialects to the
> necessary names. For driver-specific functions you could potentially
> have a prefix that indicates it should flow through into the resulting
> xml, e
The preprocessor approach might be redundant if I can
get the grammar to recognize the old syntax. Of
course, that's a medium sized 'if' there.
--- Gavin King <[EMAIL PROTECTED]> wrote:
> I think this is our plan. Any insurmountable
> problems with that, do you
> think?
>
> (Devils in the detai
LOL! Ya know what? I had implemented this *exact
same* tranform back when I was writing a 4GL compiler
at work.
This one's easy though:
ident '('[METHOD_CALL] expr => expr '.' ident
'('[METHOD_CALL]
Especially since parens are not use for array
subscripts, like they were in the 4GL. :-P
I'll
On 18 Nov (00:17), dosapati wrote:
> Is it possible to use Hibernate in long duration transactions. Where the transaction
> can be across user interactions with the system. Can span minutes, days or months.
> How can I use Hibernate session for this purpose? Do I need any additional software
> fo
On Mon, Nov 17, 2003 at 01:05:50PM -0500, Les A. Hazlewood wrote:
> > How about we get a list of ANSI-standard SQL functions and portably
> > support those and those only (I don't think there are many of them).
>
> This makes a lot of sense. However, what if something so commonly used (like a
> c
Is it possible to use Hibernate in long
duration transactions. Where the transaction
can be across user interactions with the
system. Can span minutes, days or months.
How can I use Hibernate session for
this purpose? Do I need any additional
software
for this in addition to Hibernate?
> How about we get a list of ANSI-standard SQL functions and portably
> support those and those only (I don't think there are many of them).
This makes a lot of sense. However, what if something so commonly used (like a
current_timestamp() function) is not part of the ANSI standard (assuming ANSI
I think this is our plan. Any insurmountable problems with that, do you
think?
(Devils in the details, obviously.)
We might almost be able to replace the hql package with a
preprocessor/postprocessor that predigests the query into a form that
the new parser can swallow. I don't think this is i
How about we get a list of ANSI-standard SQL functions and portably
support those and those only (I don't think there are many of them).
David Channon wrote:
I think limited application of aliases would be beneficial but query
substitutions would work in many situations. The current date issue I
breaking to much existing code. Don't want to do this lightly. Is this step
with the new parser and possible changes to the supported syntax a Version 3
option?
I think so.
No. we need this now. The longer we stay with the existing parser, the
harder it becomes to ditch it in the long run.
We
On 18 Nov (00:47), David Channon wrote:
> Postgress= now()
Thats good.
> breaking to much existing code. Don't want to do this lightly. Is this step
> with the new parser and possible changes to the supported syntax a Version 3
> option?
I think so.
--
Christian Bauer
[EMAIL PROTECTED]
I think limited application of aliases would be beneficial but query
substitutions would work in many situations. The current date issue I am not
sure that it is one of them. I am guilty of having not tried
hibernate.query.substitutions on this level. I am not sure you can use this
to deal with the
Another thought.
Would it be nice to be able to express:
upper(customer.name)
as the more object-oriented:
customer.name.upper()
How hard would this be to implement in the parser? Especially, how hard
would it be to make the parser spit out the exact same AST in either case?
--
Gavin K
Sounds like a reasonable compromise, as making the grammar a superset of all
proprietary grammars is also on the ambitious side (done this before, it was
horribly arcane and fragile... twitch-twitch). Just thinking out loud
here... maybe some form of pass-through quoting would be an elegant way to
No problem. I'll just make 'by' optional. There goes my idea about making
multi-word keywords! LOL!
Seriously though, 'having' is a sub-clause of group still, right? I could
make the nodes appear on the same level without much effort if you prefer.
I figured the writers of the 'group' analysis
I don't think we are realistically going to be able to dream up portable
aliases for everything. I can see an argument for doing it for a couple
of very important cases like current date. Beyond that I wouldn't really
like to go there. Note that you can't have both really ... as soon as
you hav
David,
Excelent point. It looks like a design decision is required here (i.e.
portable aliases vs. proprietary pass-through). Personally, I like the idea
of portable aliases for special features, as I'm not sure how I'm going to
plug Hibernate's dialect information into the parser. It's probably
Christian,
Thanks for the idea. Just tried it... new with any expr list, including
subqueries are parsed. The AST came out a little funny at first, but I fixed
that using a new 'CONSTRUCTOR' node type (first sub-node is the class,
subsequent nodes are the argument list).
input: ->select new Foo(c
Finally, why is the "having" node a submode of the "group" node?
Because it is a dependent sub-clause of 'group by' at the moment. Is that
not the case?
H. I suppose it is. well, as long as we can support
from . where group having .
as well as the traditional
from .
Responses below
> -Original Message-
> From: Gavin King [mailto:[EMAIL PROTECTED]
> Sent: Monday, November 17, 2003 6:50 AM
> > -- Grammar problem #1: The 'from Order order where' examples --
> >
> > Also, notice that I changed 'order' to 'ord' in many of the HQL
> > examples in the
On 18 Nov (00:05), Gavin King wrote:
> Does the parser handle:
>
> * subselects (can't notice one anywhere)
> * floating point and decimal literals
> (it doesn't need to parse them, just recognize them)
I'd also plan for "select new Foo(count(bar))".
> This is the "implicit group by" we discu
Does the parser handle:
* subselects (can't notice one anywhere)
* floating point and decimal literals
(it doesn't need to parse them, just recognize them)
We also want to plan for the following new syntax:
select foo, count(bar)
from Foo foo outer join foo.bars bar
group
This is the "imp
-- Grammar problem #1: The 'from Order order where' examples --
Also, notice that I changed 'order' to 'ord' in many of the HQL examples in
the hibernate documentation. This is because 'order' is a keyword, and the
statement:
Don't worry. Just outlaw keywords as identifiers. The reality is that
30 matches
Mail list logo
