you could always put a big warning in the README, and mark xmpbox/preflight as 
deprecated.
that should be sufficient for people not to use it while you work on doing the 
modification for 1.8.
On May 23, 2012, at 7:11 AM, Guillaume Bailleul wrote:

> Hi,
> 
> I agree with Andreas on that point : we should not release a code that
> will be dead soon.
> 
> preflight (and xmpbox) has never been released. So, in my opinion,
> PDFBox 1.7 can be released without preflight... it bothers me for a
> project but I think it is the best thing to do.
> 
> There is more than one issue linked with that:
> * Eric's new version will change the interface
> * xmpbox and jempbox should be merged (or at least one should desappear).
> 
> 
> The schedule with these modification is not compatible with 1.7 release.
> 
> 
> Guillaume
> 
> 
> On Tue, May 22, 2012 at 9:19 PM, Leleu Eric <[email protected]> wrote:
>> Hi,
>> 
>> 2012/5/22 Andreas Lehmkuehler <[email protected]>
>> 
>>> Hi,
>>> 
>>> Am 21.05.2012 20:38, schrieb Leleu Eric:
>>> 
>>>  Hi,
>>>> 
>>>> I discussed with Guillaume about the Preflight refactoring and we have
>>>> some
>>>> questions about le preflight module and the next release.
>>>> 
>>> Hmmm, last minute timing .... I planned to cut the release now.
>>> 
>>> 
>> I'm sorry :(
>> 
>> 
>>> 
>>>  The "new" preflight implementation will be more flexible and configurable.
>>>> But it  will  have significant impact on the current implementation. (New
>>>> interface, new way to load/validate the pdf...)
>>>> 
>>>> Due to the 1.7 release that is coming soon, we would like to know how we
>>>> should commit the preflight modifications without "breaking" the 1.7
>>>> release.
>>>> 
>>>> Here is 3 possibilities :
>>>> 
>>>> 1 - Exclude the preflight module from the release and work with the
>>>> current
>>>> code without compatibility with the old version.
>>>> 
>>> Any suggestions how to do that easily?
>> 
>> 
>> I was thinking to comment the preflight declaration in the pdfbox-reactor
>> pom file.
>> Doing this, the source code will be present in the tagged version, but they
>> will not have preflight jar in the repository with the version 1.7.
>> 
>> It's may not be the best way to achieve that but it is probably the easiest
>> way...
>> 
>> 
>>>  2 - Include the preflight module in the 1.7 release, the new
>>>> implementation
>>>> will be create in new packages. They will have no more modifications of
>>>> the
>>>> old implementation that will be marked as deprecated. Expect bug fix
>>>> nothing will be done on old version, only new implementation will be
>>>> improved (new configuration capabilities, new validation format...). Old
>>>> version will be definitely removed in a further release (1.8 or 2.0 ?)
>>>> 
>>>> 3 - Include the preflight module unchanged and release it. The new
>>>> validation tool will be done in a new module (Present only in the 2.0
>>>> branch ?). They will have no more modifications of the old module that
>>>> will
>>>> be marked as deprecated. Expect bug fix nothing will be done on old
>>>> version. Only new module will evolve (new configuration capabilities, new
>>>> validation format...).
>>>> 
>>> I don't like the idea of releasing code for the first time which is
>>> already meant to be dead.
>>> 
>> 
>> 
>>>  What is your opinion about that ?
>>>> 
>>> IMHO, as there wasn't any Apache release of preflight the cleanest way
>>> would be to replace the "old" implementation using the "new" one. But there
>>> are some questions:
>>> 
>>> - Is the "new" implementation completed? If not, when do you expect to be
>>> ready?
>>> 
>> 
>> 
>> No, it isn't. Hopefully I will finish the refactoring at the end of June
>> (but without guaranty), so it will be quite late for the 1.7 release. :(
>> 
>> 
>> - Is the "new" implementation a full replacement of the old one ?
>>> 
>> 
>> 
>> No, I thought that around 50% of the code will be reused.
>> 
>> 
>> 
>>> Depending on your answers and other opinions we might postpone the release
>>> for a couple of weeks.
>>> 
>>> 
>>>  If you have any questions, do not hesitate.
>>>> 
>>> Don't hesitate to answer fast ;-)
>>> 
>> 
>> One more time, I'm really sorry to delay the pdfbox release.
>> 
>> 
>>>  BR,
>>>> 
>>>> Eric
>>>> 
>>>> 
>>> BR
>>> Andreas Lehmkühler
>>> 
>> 
>> BR,
>> Eric

Reply via email to