9:28 a.m.
After a significant hiatus, I have restarted my work on adding improved
support for Firebird 2.5 and 3.0 (and more importantly struggling
through some test failures).
I am wondering what I should target for my changes: the current master
branch or the upstream/wip/6.0 branch. What would be the preferred or
'better' option?
Mark
--
Mark Rotteveel
8:19 a.m.
On 16-6-2019 16:28, Mark Rotteveel wrote:
After a significant hiatus, I have restarted my work on adding
improved
support for Firebird 2.5 and 3.0 (and more importantly struggling
through some test failures).
I am wondering what I should target for my changes: the current master
branch or the upstream/wip/6.0 branch. What would be the preferred or
'better' option?
On a related note, I have touched a lot of tests to either skip them or
make some changes to make them pass. What is the best approach: commit
together with the dialect changes, or offer the dialect change in a
separate pull request from all those test changes?
Mark
--
Mark Rotteveel
5:12 a.m.
On Tue, 18 Jun 2019 at 14:19, Mark Rotteveel <mark(a)lawinegevaar.nl> wrote:
On 16-6-2019 16:28, Mark Rotteveel wrote:
> After a significant hiatus, I have restarted my work on adding improved
> support for Firebird 2.5 and 3.0 (and more importantly struggling
> through some test failures).
>
> I am wondering what I should target for my changes: the current master
> branch or the upstream/wip/6.0 branch. What would be the preferred or
> 'better' option?
Hi Mark,
sorry for the late reply, I think we'll need Steve to give you a
definitive answer.
Just wondering, would your patches be significantly different?
As far as I know branch wip/6.0 is a bit in flux ATM, so it might be
safer to just target master.
Or send them to both if that's easy for you?
On a related note, I have touched a lot of tests to either skip them
or
make some changes to make them pass. What is the best approach: commit
together with the dialect changes, or offer the dialect change in a
separate pull request from all those test changes?
Either works for us. Maybe it's easier for you to first send a PR with
all testsuite improvements?
Thanks,
Sanne
Mark
--
Mark Rotteveel
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
10:09 a.m.
On 2019-06-19 12:12, Sanne Grinovero wrote:
On Tue, 18 Jun 2019 at 14:19, Mark Rotteveel
<mark(a)lawinegevaar.nl>
wrote:
>
> On 16-6-2019 16:28, Mark Rotteveel wrote:
> > After a significant hiatus, I have restarted my work on adding improved
> > support for Firebird 2.5 and 3.0 (and more importantly struggling
> > through some test failures).
> >
> > I am wondering what I should target for my changes: the current master
> > branch or the upstream/wip/6.0 branch. What would be the preferred or
> > 'better' option?
Hi Mark,
sorry for the late reply, I think we'll need Steve to give you a
definitive answer.
Just wondering, would your patches be significantly different?
At this point I have only looked at master, so I don't yet know if there
would be significant differences or not.
As far as I know branch wip/6.0 is a bit in flux ATM, so it might be
safer to just target master.
Or send them to both if that's easy for you?
I can probably do that as well.
> On a related note, I have touched a lot of tests to either skip
them
> or
> make some changes to make them pass. What is the best approach: commit
> together with the dialect changes, or offer the dialect change in a
> separate pull request from all those test changes?
Either works for us. Maybe it's easier for you to first send a PR with
all testsuite improvements?
Most of the test changes rely on the presence of the new dialects I
created for Firebird 2.5 and 3.0 (for dialect-specific checks, or to
skip the test for a dialect). A single PR seems like a logical choice,
but the new dialects could be the first PR and the test changes as a
follow up.
Mark
7:49 a.m.
On 21-6-2019 17:09, Mark Rotteveel wrote:
On 2019-06-19 12:12, Sanne Grinovero wrote:
> As far as I know branch wip/6.0 is a bit in flux ATM, so it might be
> safer to just target master.
> Or send them to both if that's easy for you?
I can probably do that as well.
I have my dialect changes for master ready and will open the pull
request shortly.
I also started on a branch with the dialects for wip/6.0 but I'm running
into all kinds of issues with query generation. Some examples:
- deletes are generated as `delete <tablename>` instead of `delete from
<tablename>` (SQL standard syntax requires `from`, as does Firebird).
This happens in SqlDeleteToJdbcDeleteConverter.processDeleteStatement
- joins add parentheses around table + alias elements (eg `inner join
(<tablename> as <alias) on ...` (if I follow the rules of 7.10 <joined
table> in SQL:2016 correctly, parentheses around table + alias are not
in the standard grammar, nor is it allowed by Firebird). This seems to
happen in AbstractSqlAstWalker.visitTableGroupJoin.
I could create a pull request for my current state on wip/6.0 and update
it later, but I wonder if it would be better to wait until wip/6.0 is in
a better state. What do you think?
Mark
--
Mark Rotteveel
8:07 a.m.
On 29-6-2019 14:49, Mark Rotteveel wrote:
On 21-6-2019 17:09, Mark Rotteveel wrote:
> On 2019-06-19 12:12, Sanne Grinovero wrote:
>> As far as I know branch wip/6.0 is a bit in flux ATM, so it might be
>> safer to just target master.
>> Or send them to both if that's easy for you?
>
> I can probably do that as well.
I have my dialect changes for master ready and will open the pull
request shortly.
PR: https://github.com/hibernate/hibernate-orm/pull/2934
--
Mark Rotteveel
2479
days inactive
2492
days old
5 comments
2 participants
participants (2)
-
Mark Rotteveel -
Sanne Grinovero