On 2/23/07, Gilles Faucherand <[EMAIL PROTECTED]> wrote:
It's done.
But, bug by comparing the ivy file of the module and the ivy file copied
in the repository,
the status attribute seems to be not correctly set ... (first line is
the ivy file module)
< <info organisation="MyCompany" module="C" revision="1.0.1"
status="release" />
---
> <info organisation="MyCompany" module="C" revision="1.0.1"
status="integration" publication="20070223151439"/>
Whereas if I use the status attribute of the publish task, it works...
Another JIRA issue ?
No, according to the documentation of the publish task, the status
defaults to ${
ivy.status
}, and is thus always set during publish, no matter the one found in
your original ivy file.
Xavier
Gilles
Gilles Faucherand wrote:
> Xavier Hanin wrote:
>> On 2/23/07, Gilles Faucherand <[EMAIL PROTECTED]> wrote:
>>>
>>> Xavier Hanin wrote:
>>> > On 2/23/07, Gilles Faucherand <[EMAIL PROTECTED]>
wrote:
>>> >>
>>> >> Hi,
>>> >>
>>> >> I have some problems with eviction strategy. I set a
>>> >> defaultConflictManager="latest-time"
>>> >> in my ivyconf meaning for me if a conflict appears to retrieve
>>> all the
>>> >> time the last in time published.
>>> >>
>>> >> Say for instance that I have 3 modules A, B and C.
>>> >> 1) I publish C-1.0.0 and later C-1.0.1.
>>> >> 2) B needs C-1.0.0, I retrieve (ok) and I publish B-1.0.0
>>> >> 3) A needs B-1.0.0 and C-1.0.1 (order is important: B before C
>>> in ivy
>>> >> file ...)
>>> >>
>>> >> When I retrieve I get C-1.0.0 ???
>>> >> [resolve] :: evicted modules:
>>> >> [resolve] [ MyCompany | C | 1.0.1 ] by [[ MyCompany | C |
>>> 1.0.0]]
>>> >> in [all]
>>> >> [resolve] in [ MyCompany | A | 1.0.0 ] with latest-time
>>> >>
>>> >> If I change the order in the ivy file i.e. C before B, I get what I
>>> want
>>> >> C-1.0.1 ...
>>> >> [resolve] :: evicted modules:
>>> >> [resolve] [ MyCompany | C | 1.0.0 ] by [[ MyCompany | C |
>>> 1.0.1]]
>>> >> in [all]
>>> >> [resolve] in [ MyCompany | B | 1.0.0 ] with latest-time
>>> >>
>>> >> Is it normal ?
>>> >
>>> >
>>> > No, this is not normal. The only reason why this should happen is if
>>> > the two
>>> > revision are considered equal according to the latest revision
>>> strategy
>>> > used, in your case latest-time. So this should only happen if the
two
>>> > have
>>> > the same time of publication. Note that Ivy uses the publication
time
>>> > in the
>>> > Ivy file during conflict resolution, so make sure you have set the
>>> > publication attribute in your ivy files in your repository.
>>> You mean in the ivy file of the module C <info
>>> organisation="MyCompany"
>>> module="C" ... publication="xxxxx" >
>>> No, or I should change this date each time I publish and I've got a
>>> nightly builder ...
>>>
>>> After publication in the repository the ivy file created was
>>> myrepos/MyCompany/C/ivy-1.0.0.xml (following my ivy artifact pattern)
>>>
>>> <?xml version="1.0" encoding="UTF-8"?>
>>> <ivy-module version="1.0">
>>> <info organisation="MyCompany" module="C" revision="1.0.0"
>>> status="integration" publication="20070223132315"/>
>>> ...
>>> </ivy-module>
>>>
>>> A publication date is set ... I thought this date was used ?
>>
>>
>> Yes, this the one which should be used: the ivy file for C 1.0.0 and
>> the one
>> for C 1.0.1 in your repository. Compare the two publication dates and
>> make
>> sure they are different.
> yes, they are:
>
> < <info organisation="MyCompany" module="C" revision="1.0.0"
> status="integration" publication="20070223142657"/>
> ---
> > <info organisation="MyCompany" module="C" revision="1.0.1"
> status="integration" publication="20070223142715"/>
>
> ok, I add an issue in JIRA.
>
> Thanks for your help.
>
> Gilles
>
>> But I realize that during conflict resolution, the
>> ivy files are not downloaded, so it seems to be a bug in Ivy to try
>> to rely
>> on this publication date in the ivy file. It should better rely on
>> the last
>> modified timestamp of the file. Could you please add an issue in JIRA?
>>
>> - Xavier
>>
>>
>>> If you still
>>> > have the problem, you can open a JIRA issue describing your problem
>>> > and your
>>> > environment.
>>> >
>>> > Xavier
>>> >
>>> >
>>> >
>>> > Thanks in advance,
>>> >> Gilles
>>> >>
>>> >>
>>> >
>>>
>>>
>>
>
>
