On Sun, Jan 6, 2013 at 4:21 PM, Niall Pemberton
<niall.pember...@gmail.com> wrote:
> On Wed, Dec 26, 2012 at 6:32 PM, Phil Steitz <phil.ste...@gmail.com> wrote:
>> On 12/24/12 1:13 PM, Gilles Sadowski wrote:
>>> On Mon, Dec 24, 2012 at 09:27:57AM -0800, Phil Steitz wrote:
>>>> On 12/24/12 4:37 AM, Gilles Sadowski wrote:
>>>>> Hi.
>>>>>
>>>>>> [...]
>>>>>>
>>>>>> Commons Math can be downloaded from the following page:
>>>>>>   http://commons.apache.org/math/download_math.cgi
>>>>> I missed that the "mvn commons:download-page" had generated a new
>>>>> template page:
>>>>>   src/site/xdoc/download_math3.xml
>>>>> instead of modifying
>>>>>   src/site/xdoc/download_math.xml
>>>>>
>>>>> Questions:
>>>>> 1. Is the creation of a new template page expected?
>>>>> 2. In the affirmative, is it automatically taken into account by the 
>>>>> "site"
>>>>>    generation? I.e. which of the old and new template will eventually be
>>>>>    used to generate the HTML page?
>>>>> 3. Should the old template be deleted from SVN?
>>>>> 4. Given that I did not notice the creation of the new template, it was 
>>>>> not
>>>>>    included in the tag, and the download page on the site misses the links
>>>>>    to the new release files. How to fix that?
>>>> I think I understand what is going on and how to "fix" it, but am
>>>> not 100% sure.
>>>>
>>>> I think the root cause of the problem is r1392022 of the [math] pom,
>>>> where we changed the commons.component.id property to get correct
>>>> osgi bundles created.
>
> Yes, but I think it would have been better to leave the
> commons.component.id property unchanged and add a
> commons.osgi.symbolicName property to commons math's pom to get what
> you wanted for the osgi bundle.

I have attached a patch to MATH-876 which does this - so should
resolve all issues

https://issues.apache.org/jira/browse/MATH-876

> Niall
>
>>>> This causes the download plugin to generate
>>>> the second template above.  The site plugin does create an html
>>>> page, so there are two.  What I don't get is why only one of them
>>>> ends up on p.a.o (the "old" one).  In any case, the "new" one has
>>>> the wrong name, so this needs to be fixed.
>>>>
>>>> I don't know enough about the download plugin to figure out how to
>>>> really fix this.  Here is a temporary hack that should fix things.
>>>>
>>>> 0) Change commons.component.id back to "math" locally in the pom.
>>>> 1) Change commons.release.version back to "3.1"
>>>> 2) run commons:download
>>>> 3) check in the modified template.
>>>> 4) generate the site locally
>>>> 5) scp just the download page (download_math.html) to
>>>> p.a.o/www/commons.apache.org/math
>>>>
>>>> It would probably also work to undo the pom changes after step 3)
>>>> and then do a full site build and deploy.
>>> Thanks but, hack for hack, I took the more direct route to directly modify
>>> the HTML page (which is equivalent to your step 5, IIUC). I hope there
>>> aren't any ill side effects...
>>
>> That was actually my first thought ;)
>>>
>>> The rest should be really fixed, i.e. keeping the correct component id and
>>> having the plugin generate/modify the correct file.
>>>
>>> Perhaps, we should open a JIRA report in order not to forget.
>>
>> Agreed, but I am not sure if it should be against [math],
>> commons-parent or the download plugin.
>>
>> It looks like the download plugin and the osgi plugin both use the
>> property commons.component.id.  The value of this property ends up
>> embedded in the template name - download_xxx.html - that the
>> download plugin generates.  This name *must* match download_xxx.cgi
>> and the link in site.xml for the download pages to work.  I guess
>> one way we can fix this is to rename download_math.cgi (and fix the
>> links that point to it elsewhere); but that seems wrong to me.
>> Maybe the best approach is to define commons.project.id and have the
>> download plugin use that property instead.  What do others think?
>>
>> One other question I have is where exactly does download_math.cgi
>> come from?  I don't see it generated locally or in svn anywhere.
>>
>> Phil
>>>
>>>
>>> Regards (and Merry Christmas!),
>>> Gilles
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
>>> For additional commands, e-mail: dev-h...@commons.apache.org
>>>
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
>> For additional commands, e-mail: dev-h...@commons.apache.org
>>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org

Reply via email to