continuum-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Marcelo Fukushima" <takesh...@gmail.com>
Subject Re: continuum-store and JDO
Date Fri, 12 Jan 2007 11:22:15 GMT
On 1/12/07, Rahul Thakur <rahul.thakur.xdev@gmail.com> wrote:
>
> Marcelo Fukushima wrote:
> > On 1/11/07, Rahul Thakur <rahul.thakur.xdev@gmail.com> wrote:
> >>
> >>
> >> Marcelo Fukushima wrote:
> >> > yeah...
> >> > but now that ive settled it, ive encountered a new set of probs, this
> >> > time in the data-management with the trunk on svn:
> >> > -while backing up the continuum store, a FileWriter is used (wich uses
> >> > default system char encoding), but the stream used to write the xml
> >> > tries to use utf-8, in wich case an exception is thrown;
> >>
> >> This is a known issue. I think its to do with locale settings.
> >
> > ive attached a patch that i think solves this one (it actually uses
> > the same charset encoding in the writer by using a OutputStreamWriter
> > ( FileOutputStream, Charset )
> >
> > i know this test passes to me now but itd be great if anyone else
> > could test it in a diff env (maybe other jdks and os's)
> >
> > http://jira.codehaus.org/secure/ManageAttachments.jspa?id=45956
>
> Thanks! I will take it for a spin (hopefully shortly) though i am on
> Windows XP as well. Yep, it might be a good idea if we can get it to run
> on another machine as well.
>
> >
> >>
> >> > -the backup file (an xml) is compared to a hard coded xml file... but
> >> > at least with my dev env (win xp with jdk6), some child nodes are
> >> > swaped...
> >> You mean the nodes _do_ exist _but_ in a different order.  Can you point
> >> me to the test that is showing this behaviour?
> >
> > exactly... ill generate it again, but i think the id tag gets
> > switched... its in the DataManagementToolTest inside the
> > continuum-data-management module
>
> Cool, will have a look when I get back home. I came across a different
> behaviour with HashMap key ordering in JDK 6.  I suspect if its
> something similar in this case.

probably yeah... should i raise a jira issue? the fix is to compare
the DOM structure of the xmls instead of the string representation

>
> >
> > btw i cant seen to compile from maven: it gives me an Access Denied
> > Exception on some random directories (im compiling and running the
> > tests from inside eclipse)
>
> Do you have continuum checkout on your drive's root or nested a few
> levels in another directory ?

yup in the root of my driver... the weird thing is: the problem occurs
totally randomly and sometimes not at all... it happens while
exploding the war and only if i run tasks that package it from the
continuum-parent - it doesnt happen when i run from inside the webapp
module

its very confusing to me, but i suspect it is related to this messy filesystem

>
> Rahul
>
> >
> >>
> >> Cheers,
> >> Rahul
> >>
> >> >
> >> > the first one is most certainly a bug (and i already have a fix),
> >> > while the second one im not so sure so i wanted to ask yall first
> >> >
> >> > regards,
> >> > takeshi
> >> >
> >> > On 1/11/07, Emmanuel Venisse <emmanuel@venisse.net> wrote:
> >> >> windows path length pb?
> >> >>
> >> >> Marcelo Fukushima a écrit :
> >> >> > im looking into it, but i cant seen to install the modules
> >> locally -
> >> >> > im getting a weird random access denied error...
> >> >> >
> >> >> >
> >> >> > On 1/10/07, Emmanuel Venisse <emmanuel@venisse.net> wrote:
> >> >> >> Yes. A global patch would be better instead of separate patches.
> >> >> >> I tried your 2 patches and updated locally jpox-* to 1.1.3
in
> >> >> >> continuum parent pom. With them, all tests works fine, but
> >> continuum
> >> >> >> doesn't start.
> >> >> >>
> >> >> >> Emmanuel
> >> >> >>
> >> >> >> Marcelo Fukushima a écrit :
> >> >> >> > Hello dear devs.
> >> >> >> >
> >> >> >> > After everyone's help, ive noticed that jpox was saving
the
> >> >> >> > BuildDefinition with wrong values (the buildFresh and
> >> arguments and
> >> >> >> > possibly others too)... since i dont know how jpox work,
the
> >> first
> >> >> >> > thing i tried was to update to a more recent version
of it
> >> (namely
> >> >> >> > 1.1.3) and everything worked fine...
> >> >> >> > im going to attach the test case to jira (actually a
patch to
> >> the
> >> >> >> > testcase to check those things too), but should i attach
a
> >> patch to
> >> >> >> > the pom too?
> >> >> >> >
> >> >> >> > On 1/9/07, Brett Porter <brett@apache.org> wrote:
> >> >> >> >> The metadata file (and enhanced classes) are generated
when you
> >> >> use
> >> >> >> >> Maven to build the project. As long as eclipse doesn't
> >> >> overwrite the
> >> >> >> >> class files it should be fine. Alternatively, you
can use the
> >> >> eclipse
> >> >> >> >> plugin for jpox to make sure the classes get enhanced
and
> >> just run
> >> >> >> >> maven once to generate the metadata file. I've never
used that
> >> >> >> >> plugin, but it is available from the jpox site.
> >> >> >> >>
> >> >> >> >> Thanks!
> >> >> >> >> - Brett
> >> >> >> >>
> >> >> >> >> On 09/01/2007, at 2:27 PM, Marcelo Fukushima wrote:
> >> >> >> >>
> >> >> >> >> > hello folks! i sent a couple of emails to the
user list,
> >> but i
> >> >> >> guess i
> >> >> >> >> > could help a little too, right? so i just checked
out the
> >> >> code from
> >> >> >> >> > SVN and wanted to tackle
> >> >> >> >> > http://jira.codehaus.org/browse/CONTINUUM-1103
> >> >> >> >> > while i could isolate the bug (the property
is not getting
> >> >> persisted
> >> >> >> >> > on the db) but since i know almost nothing about
jdo, i cant
> >> >> run the
> >> >> >> >> > tests inside eclipse to try to figure out the
solution and i
> >> >> couldnt
> >> >> >> >> > find where the metadata file or the enhanced
version of the
> >> >> classes
> >> >> >> >> > are found...
> >> >> >> >> > is there any doc that can help me with that
kinda of info?
> >> >> thanks
> >> >> >> >> > in advance
> >> >> >> >> > --
> >> >> >> >> > []'s
> >> >> >> >> > Marcelo Takeshi Fukushima
> >> >> >> >>
> >> >> >> >
> >> >> >> >
> >> >> >>
> >> >> >>
> >> >> >
> >> >> >
> >> >>
> >> >>
> >> >
> >> >
> >>
> >
> >
>


-- 
[]'s
Marcelo Takeshi Fukushima

Mime
View raw message