You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@opennlp.apache.org by Grant Ingersoll <gs...@apache.org> on 2010/12/02 23:59:04 UTC
OpenNLP Site
I would suggest we setup and use the new Apache CMS for our site. See http://www.apache.org/dev/cms.html
Thoughts? I can put in the request and get the initial stuff setup.
-Grant
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 12:45 AM, James Kosin wrote:
> +1 for Jeorn's current suggestion; which makes things easier.
>
Benson pointed out that my proposed layout does not
leave space for additional things we might add in the future,
like a sandbox or a project for training data, regression testing
project, etc.
As I said, I agree with his view. He proposed a layout
which leaves room for future additions.
Jörn
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 6:23 PM, Benson Margulies wrote:
> On Thu, Dec 2, 2010 at 6:18 PM, James Kosin <ja...@gmail.com> wrote:
>> On 12/2/2010 6:10 PM, Jörn Kottmann wrote:
>>> No that is the correct location:
>>> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3C4CF81406.3060103@gmail.com%3E
>>>
>>>
>>> Jörn
>>>
>> I thought svn took care of the branch and trunk directories. They are
>> structural directories that get created when you commit. Then branches
>> will contain all branches and the trunk gets updated. Tags also get
>> their own directories.
> Nope. Nothing is automatic. TTB is a convention, you have to create
> those dirs in svn. If you release with maven, it will automatically
> make tags, but nothing will make them.
>
Thanks, I see in the book I'm reading. There are two general layouts.
One like Jeorn is proposing and another with the trunk/tags/branches
under each sub-project of the whole.
+1 for Jeorn's current suggestion; which makes things easier.
James
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
Ooops,
my previous email should probably have read:
/opennlp
/trunk
/opennlp
/tools
/uima
/maxent
/branches
/tags
....
Re: OpenNLP Site
Posted by Jason Baldridge <jb...@mail.utexas.edu>.
I think it is important to have a separate maxent release -- there are many
people who use that package without using the OpenNLP toolkit.
On Thu, Dec 2, 2010 at 6:45 PM, Jörn Kottmann <ko...@gmail.com> wrote:
> Historically all three parts have been released independent
> of each other. The Maxent version and OpenNLP version
> do not match.
>
> In my opinion we should release everything together and ship
> it in one package in the future. The Tools and Uima version is
> already aligned, after the big maxent refactoring the maxent version
> can also be aligned.
>
> Our first release at Apache could be one package, which contains
> all three artifacts in these versions:
> OpenNLP Tools 1.5.1
> OpenNLP UIMA 1.5.1
> OpenNLP Maxent 3.0.1
>
> Jörn
>
>
> On 12/3/10 1:35 AM, Benson Margulies wrote:
>
>> James,
>>
>> It all depends on your proposed release architecture. Every releasable
>> thing, in my opinion, gets its own TTB.
>>
>> That's the idea of the /products dir in the proposal I sent.
>>
>> --benson
>>
>>
>> On Thu, Dec 2, 2010 at 7:13 PM, James Kosin<ja...@gmail.com>
>> wrote:
>>
>>> Hmmm...
>>>
>>> Maybe this would be better:
>>>
>>> /opennlp
>>> /tools
>>> /tags
>>> /branches
>>> /trunk
>>> /uima
>>> /tags
>>> /branches
>>> /trunk
>>> /maxent
>>> /tags
>>> /branches
>>> /trunk
>>>
>>> I know it looks odd, but, when we decide to move maxent on to the other
>>> project, It won't be as destructive and allows us to keep the three a
>>> bit separated.
>>> What is Apache's requirement on releases? Do they have to be a single
>>> tar? Or can it be several independent library modules?
>>>
>>> This way we can still have:
>>> opennlp-maxent-incubator-.......
>>> opennlp-uima-incubator-......
>>> opennlp-tools-incubator-......
>>>
>>> James
>>>
>>>
>
--
Jason Baldridge
Assistant Professor, Department of Linguistics
The University of Texas at Austin
http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 7:45 PM, Jörn Kottmann wrote:
> Historically all three parts have been released independent
> of each other. The Maxent version and OpenNLP version
> do not match.
>
> In my opinion we should release everything together and ship
> it in one package in the future. The Tools and Uima version is
> already aligned, after the big maxent refactoring the maxent version
> can also be aligned.
>
> Our first release at Apache could be one package, which contains
> all three artifacts in these versions:
> OpenNLP Tools 1.5.1
> OpenNLP UIMA 1.5.1
> OpenNLP Maxent 3.0.1
>
> Jörn
>
If we are going to align everything and release together then a
variation on this architecture may work:
/opennlp
/trunk
/tools
/uima
/maxent
/branches
/tags
/sandbox
...
James
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
Historically all three parts have been released independent
of each other. The Maxent version and OpenNLP version
do not match.
In my opinion we should release everything together and ship
it in one package in the future. The Tools and Uima version is
already aligned, after the big maxent refactoring the maxent version
can also be aligned.
Our first release at Apache could be one package, which contains
all three artifacts in these versions:
OpenNLP Tools 1.5.1
OpenNLP UIMA 1.5.1
OpenNLP Maxent 3.0.1
Jörn
On 12/3/10 1:35 AM, Benson Margulies wrote:
> James,
>
> It all depends on your proposed release architecture. Every releasable
> thing, in my opinion, gets its own TTB.
>
> That's the idea of the /products dir in the proposal I sent.
>
> --benson
>
>
> On Thu, Dec 2, 2010 at 7:13 PM, James Kosin<ja...@gmail.com> wrote:
>> Hmmm...
>>
>> Maybe this would be better:
>>
>> /opennlp
>> /tools
>> /tags
>> /branches
>> /trunk
>> /uima
>> /tags
>> /branches
>> /trunk
>> /maxent
>> /tags
>> /branches
>> /trunk
>>
>> I know it looks odd, but, when we decide to move maxent on to the other
>> project, It won't be as destructive and allows us to keep the three a
>> bit separated.
>> What is Apache's requirement on releases? Do they have to be a single
>> tar? Or can it be several independent library modules?
>>
>> This way we can still have:
>> opennlp-maxent-incubator-.......
>> opennlp-uima-incubator-......
>> opennlp-tools-incubator-......
>>
>> James
>>
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
svn doesn't know anything about it :-) However, maven does.
People type 'svn cp trunk tags/my-release-4' to make a release tag.
If you build with maven, the release plugin will do this with no fussy
config if the name is 'tags'. If it's 'tag', you have to tell Maven
about it.
For you new-to-svn types, the most important thing to know for now is
that a svn tag is just like a svn branch. Both happen by running a
'copy' operation.
On Thu, Dec 2, 2010 at 8:41 PM, James Kosin <ja...@gmail.com> wrote:
> On 12/2/2010 8:36 PM, Jörn Kottmann wrote:
>> On 12/3/10 2:20 AM, James Kosin wrote:
>>> Thanks,
>>> Benson, that should be a /tags directory. /tag wouldn't work....
>>>
>>> Other than that it looks good.
>>>
>>> James
>>
>> /tag actual does work, since its really only a convention
>> (its usually a best practice to follow the convention).
>> Its even possible to use subversion without trunk, tags
>> and branch. Like we might do with the site folder.
>>
>> Jörn
> Actually, when svn performs a tag, its expecting the /tags directory.
>
> Unless svn can be configured to use tag instead of tags for the structure?
>
> James
>
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 8:36 PM, Jörn Kottmann wrote:
> On 12/3/10 2:20 AM, James Kosin wrote:
>> Thanks,
>> Benson, that should be a /tags directory. /tag wouldn't work....
>>
>> Other than that it looks good.
>>
>> James
>
> /tag actual does work, since its really only a convention
> (its usually a best practice to follow the convention).
> Its even possible to use subversion without trunk, tags
> and branch. Like we might do with the site folder.
>
> Jörn
Actually, when svn performs a tag, its expecting the /tags directory.
Unless svn can be configured to use tag instead of tags for the structure?
James
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 1:21 PM, Thilo Goetz wrote:
> On 12/3/2010 13:08, Benson Margulies wrote:
>>>
>>> I don't buy into that argument. I see no problem with
>>
>> git-svn required extra options to deal with this, as does the
>> maven-release-plugin. It's a convenience thing. I've seen it done both
>> ways, and in part experience this way was relatively chaotic.
>
> I agree, it's a convenience thing. The question is,
> who do you want to make it convenient for. Hopefully
> the only guy that has to battle with the maven release
> plugin is the release manager. Checking out the code
> is something that all developers and a lot of serious
> users do all the time. Don't know about git, but I
> would be seriously disappointed if this (to my mind
> standard) svn layout threw a spanner in the git-svn
> works.
I use to work a lot on the command line, having three
different trunks mean, I have to go into each and type
svn update three times in the end to get the latest code.
Instead of typing it just once.
Jörn
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 13:08, Benson Margulies wrote:
>>
>> I don't buy into that argument. I see no problem with
>
> git-svn required extra options to deal with this, as does the
> maven-release-plugin. It's a convenience thing. I've seen it done both
> ways, and in part experience this way was relatively chaotic.
I agree, it's a convenience thing. The question is,
who do you want to make it convenient for. Hopefully
the only guy that has to battle with the maven release
plugin is the release manager. Checking out the code
is something that all developers and a lot of serious
users do all the time. Don't know about git, but I
would be seriously disappointed if this (to my mind
standard) svn layout threw a spanner in the git-svn
works.
--Thilo
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
>
> I don't buy into that argument. I see no problem with
git-svn required extra options to deal with this, as does the
maven-release-plugin. It's a convenience thing. I've seen it done both
ways, and in part experience this way was relatively chaotic.
>
> trunk
> proj1
> proj2
> proj3
> branch
> proj1
> proj1branch1
> ...
> proj2
> tags
> proj1
>
> To my mind, it's totally a matter of taste. And I like
> this better ;-) My main reason would be that it provides
> a central trunk (which is what most everybody needs most
> of the time). You can just check out everything with
> svn co http://..../trunk opennlp, cd into opennlp and
> build. You don't have to go to 3 different places to
> get maxent, the toolkit and the UIMA wrappers.
>
> BTW, we have the kind of setup that Benson suggests in
> UIMA, and I find it annoying. So I speak from experience.
> Of course, as I say above, it's pretty much a matter of
> taste, and I'll go with the majority.
>
> --Thilo
>
>>
>>
>> On Thu, Dec 2, 2010 at 8:36 PM, Jörn Kottmann<ko...@gmail.com> wrote:
>>>
>>> On 12/3/10 2:20 AM, James Kosin wrote:
>>>>
>>>> Thanks,
>>>> Benson, that should be a /tags directory. /tag wouldn't work....
>>>>
>>>> Other than that it looks good.
>>>>
>>>> James
>>>
>>> /tag actual does work, since its really only a convention
>>> (its usually a best practice to follow the convention).
>>> Its even possible to use subversion without trunk, tags
>>> and branch. Like we might do with the site folder.
>>>
>>> Jörn
>>>
>
>
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 2:04 PM, Benson Margulies wrote:
> On the one hand, there are tight limits on what you can put in the source tree.
>
> On the other hand, in many legal cases, there is no such thing as GPL
> data. It is quite possible that the objectionable properties of the
> GPL will be deemed irrelevant. You would want to open a ticket on
> LEGAL on JIRA and specify the copyright information, the license, and
> exactly what was going to happen to the data.
Thanks for this advice will do this later. Looks like we will
be able to have at least some training data in our repository.
First we should proceed with agreeing on a basic svn layout ... so we
can continue
creating our new initial web site which is something I would like to get
done as
quickly as possible.
When I sent the request to Infra we have to say where the
site will be located.
Would this be a place where we can start the site:
http://svn.apache.org/repos/asf/incubator/opennlp/site/
I think its unlikely that we will get more sites anytime soon, right ?
And if so, we could simply change it.
Jörn
Re: OpenNLP Site
Posted by Grant Ingersoll <gs...@apache.org>.
On Dec 5, 2010, at 6:11 AM, Tommaso Teofili wrote:
> Hi Thilo,
>
> 2010/12/5 Thilo Goetz <tw...@gmx.de>
>
>> I've been wondering if there is something we could do
>> to enhance collaboration between all the text related
>> projects at Apache. For example, I used to be subscribed
>> to both the Solr and the Mahout lists. Eventually though,
>> I just couldn't cope with the traffic. Maybe some sort
>> of text-interest@a.o mailing list? At some point I thought
>> the Lucene umbrella could be such a focal point, but the
>> ASF is moving away from umbrellas like that. A low
>> overhead mailing list could be the ticket, it's not a
>> lot of effort, and if it doesn't take off, we just shelve
>> it. WDYT?
>>
>>
> I think that makes a lot of sense and I also wonder what can be the right
> and useful way to build such a community.
> As you say umbrella projects are not the way ASF is going to, maybe
> text-interest ML can be a good starting point.
In my experience, unfortunately, more mailing lists only exacerbate the problem, not lessen it.
Several of us are involved in all of these projects, I'd say let's get off the ground here and get used to the ASF ways and get a release under our belt, then we can collaborate more.
As for Open Relevance (ORP), there are three areas of interest:
1. IP cleared corpora
2. Judgments
3. Drivers (queries, etc.)
ORP started out of a need for Lucene, but it has always had a broader goal of supporting other projects (Mahout was my intent from day one, for instance)
As for Solr, Drew Farris, Tom and I have code that integrates OpenNLP into Solr for our book and I think the plan always has been to use that as a basis for production ready version. That being said, Tommaso recently put up a UIMA patch for Solr, so we could just leverage OpenNLP via it's UIMA bindings too.
-Grant
Re: OpenNLP Site
Posted by Jason Baldridge <jb...@mail.utexas.edu>.
That sounds good. I'm currently arranging with distributors of various
corpora to allow models trained on their materials to be used with OpenNLP
under the ASL license, but it would be great to have annotated texts that
are fully open, both for expansion/modification and for use as training
material.
On Sun, Dec 5, 2010 at 5:11 AM, Tommaso Teofili
<to...@gmail.com>wrote:
> Hi Thilo,
>
> 2010/12/5 Thilo Goetz <tw...@gmx.de>
>
> > I've been wondering if there is something we could do
> > to enhance collaboration between all the text related
> > projects at Apache. For example, I used to be subscribed
> > to both the Solr and the Mahout lists. Eventually though,
> > I just couldn't cope with the traffic. Maybe some sort
> > of text-interest@a.o mailing list? At some point I thought
> > the Lucene umbrella could be such a focal point, but the
> > ASF is moving away from umbrellas like that. A low
> > overhead mailing list could be the ticket, it's not a
> > lot of effort, and if it doesn't take off, we just shelve
> > it. WDYT?
> >
> >
> I think that makes a lot of sense and I also wonder what can be the right
> and useful way to build such a community.
> As you say umbrella projects are not the way ASF is going to, maybe
> text-interest ML can be a good starting point.
> Surely I am +1 for that.
> Cheers,
> Tommaso
>
> [1] : http://incubator.markmail.org/thread/qpzfmztgahwwa6fr
>
--
Jason Baldridge
Assistant Professor, Department of Linguistics
The University of Texas at Austin
http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by Tommaso Teofili <to...@gmail.com>.
Hi Thilo,
2010/12/5 Thilo Goetz <tw...@gmx.de>
> I've been wondering if there is something we could do
> to enhance collaboration between all the text related
> projects at Apache. For example, I used to be subscribed
> to both the Solr and the Mahout lists. Eventually though,
> I just couldn't cope with the traffic. Maybe some sort
> of text-interest@a.o mailing list? At some point I thought
> the Lucene umbrella could be such a focal point, but the
> ASF is moving away from umbrellas like that. A low
> overhead mailing list could be the ticket, it's not a
> lot of effort, and if it doesn't take off, we just shelve
> it. WDYT?
>
>
I think that makes a lot of sense and I also wonder what can be the right
and useful way to build such a community.
As you say umbrella projects are not the way ASF is going to, maybe
text-interest ML can be a good starting point.
Surely I am +1 for that.
Cheers,
Tommaso
[1] : http://incubator.markmail.org/thread/qpzfmztgahwwa6fr
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/5/2010 09:59, Isabel Drost wrote:
> On 03.12.2010 Thilo Goetz wrote:
>> Thanks, we should keep it in the backs of our
>> heads. Although we'd need different kinds of annotations,
>> we could still use the same basic data.
>
> Hmm - I'm wondering whether why the people over on open relevance shouldn't be
> open to supporting your kinds of annotations? Or are they technically so
> different from those in the mentioned use cases (IR, Machine Learning and NLP)
> that it would just make not sense to explore areas for working together?
>
>
> Isabel
I only meant that the immediate focus of the OpenRelevance
project seems to be search relevance judgments, which
is a very different problem from, say, sentence segmentation,
entity extraction or parsing. Of course we should work
together, and I have no doubts that the OpenRelevance
guys would accommodate our needs.
I've been wondering if there is something we could do
to enhance collaboration between all the text related
projects at Apache. For example, I used to be subscribed
to both the Solr and the Mahout lists. Eventually though,
I just couldn't cope with the traffic. Maybe some sort
of text-interest@a.o mailing list? At some point I thought
the Lucene umbrella could be such a focal point, but the
ASF is moving away from umbrellas like that. A low
overhead mailing list could be the ticket, it's not a
lot of effort, and if it doesn't take off, we just shelve
it. WDYT?
--Thilo
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/5/10 9:59 AM, Isabel Drost wrote:
> On 03.12.2010 Thilo Goetz wrote:
>> Thanks, we should keep it in the backs of our
>> heads. Although we'd need different kinds of annotations,
>> we could still use the same basic data.
> Hmm - I'm wondering whether why the people over on open relevance shouldn't be
> open to supporting your kinds of annotations? Or are they technically so
> different from those in the mentioned use cases (IR, Machine Learning and NLP)
> that it would just make not sense to explore areas for working together?
I also had a few ideas how such a corpus project could be done. Maybe
we should write a proposal in the wiki, and then see how we
can collaborate with the open relevance people.
Jörn
Re: OpenNLP Site
Posted by Isabel Drost <is...@apache.org>.
On 03.12.2010 Thilo Goetz wrote:
> Thanks, we should keep it in the backs of our
> heads. Although we'd need different kinds of annotations,
> we could still use the same basic data.
Hmm - I'm wondering whether why the people over on open relevance shouldn't be
open to supporting your kinds of annotations? Or are they technically so
different from those in the mentioned use cases (IR, Machine Learning and NLP)
that it would just make not sense to explore areas for working together?
Isabel
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 16:58, Isabel Drost wrote:
> On Fri, 03 Dec 2010 Thilo Goetz<tw...@gmx.de> wrote:
>> Exactly. I would really like to create some training data
>> that's under a permissive license. It's surprising how
>> little of that there is in general. It would be a lot of
>> work, but maybe we'll find help.
>
> If you are interested in gathering training data sets under permissive
> licences you may want to have a look into the Lucene Open Relevance
> project: http://lucene.apache.org/openrelevance/
>
> From the project explanation: "The Open Relevance Project (ORP) is a
> new Apache Lucene sub-project aimed at making materials for doing
> relevance testing for Information Retrieval (IR), Machine Learning and
> Natural Language Processing (NLP) into open source."
>
> Isabel
Thanks, we should keep it in the backs of our
heads. Although we'd need different kinds of annotations,
we could still use the same basic data.
--Thilo
Re: OpenNLP Site
Posted by Isabel Drost <is...@apache.org>.
On Fri, 03 Dec 2010 Thilo Goetz <tw...@gmx.de> wrote:
> Exactly. I would really like to create some training data
> that's under a permissive license. It's surprising how
> little of that there is in general. It would be a lot of
> work, but maybe we'll find help.
If you are interested in gathering training data sets under permissive
licences you may want to have a look into the Lucene Open Relevance
project: http://lucene.apache.org/openrelevance/
From the project explanation: "The Open Relevance Project (ORP) is a
new Apache Lucene sub-project aimed at making materials for doing
relevance testing for Information Retrieval (IR), Machine Learning and
Natural Language Processing (NLP) into open source."
Isabel
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 14:23, Jörn Kottmann wrote:
> On 12/3/10 2:17 PM, Thilo Goetz wrote:
>> On 12/3/2010 14:04, Benson Margulies wrote:
>>> On the one hand, there are tight limits on what you can put in the
>>> source tree.
>>>
>>> On the other hand, in many legal cases, there is no such thing as GPL
>>> data. It is quite possible that the objectionable properties of the
>>> GPL will be deemed irrelevant. You would want to open a ticket on
>>> LEGAL on JIRA and specify the copyright information, the license, and
>>> exactly what was going to happen to the data.
>>
>> +1
>>
>> Benson, I could probably find this somewhere, but do you
>> know off-hand what the ASF's position with respect to the
>> Creative Commons license is?
>
> Its here:
> http://www.apache.org/legal/resolved.html#cc-sa
>
> Seems that it is allowed to include CCA Share Alike data,
> thats BTW what wikinews is using. Could be a nice source
> for training data.
>
> Jörn
Exactly. I would really like to create some training data
that's under a permissive license. It's surprising how
little of that there is in general. It would be a lot of
work, but maybe we'll find help.
--Thilo
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 2:17 PM, Thilo Goetz wrote:
> On 12/3/2010 14:04, Benson Margulies wrote:
>> On the one hand, there are tight limits on what you can put in the
>> source tree.
>>
>> On the other hand, in many legal cases, there is no such thing as GPL
>> data. It is quite possible that the objectionable properties of the
>> GPL will be deemed irrelevant. You would want to open a ticket on
>> LEGAL on JIRA and specify the copyright information, the license, and
>> exactly what was going to happen to the data.
>
> +1
>
> Benson, I could probably find this somewhere, but do you
> know off-hand what the ASF's position with respect to the
> Creative Commons license is?
Its here:
http://www.apache.org/legal/resolved.html#cc-sa
Seems that it is allowed to include CCA Share Alike data,
thats BTW what wikinews is using. Could be a nice source
for training data.
Jörn
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 14:04, Benson Margulies wrote:
> On the one hand, there are tight limits on what you can put in the source tree.
>
> On the other hand, in many legal cases, there is no such thing as GPL
> data. It is quite possible that the objectionable properties of the
> GPL will be deemed irrelevant. You would want to open a ticket on
> LEGAL on JIRA and specify the copyright information, the license, and
> exactly what was going to happen to the data.
+1
Benson, I could probably find this somewhere, but do you
know off-hand what the ASF's position with respect to the
Creative Commons license is?
--Thilo
>
>
> On Fri, Dec 3, 2010 at 7:39 AM, Jörn Kottmann<ko...@gmail.com> wrote:
>> On 12/3/10 1:29 PM, Thilo Goetz wrote:
>>>
>>> On 12/3/2010 13:14, Jörn Kottmann wrote:
>>> [...]
>>>>
>>>> Having room for additional things like a corpus project could be nice,
>>>> because people who are interested in the source code might not
>>>> want to check out the corpus (if we ever get it created).
>>>
>>> So you can either make it convenient to check out
>>> everything, or to check out individual stuff. When
>>> I check out an open source project I tend to get
>>> everything and then later figure out if there is
>>> stuff in there that I don't need. Maybe most people
>>> are different, no idea.
>>
>> Do you know if it would be possible to check in GPL training?
>> We cannot distribute it, but we could distribute models built
>> from it.
>>
>> There is a private training data repository, but it takes ages
>> to check it out, so that might be something we should not
>> stuff into the trunk which contains the source code.
>>
>> Jörn
>>
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
On the one hand, there are tight limits on what you can put in the source tree.
On the other hand, in many legal cases, there is no such thing as GPL
data. It is quite possible that the objectionable properties of the
GPL will be deemed irrelevant. You would want to open a ticket on
LEGAL on JIRA and specify the copyright information, the license, and
exactly what was going to happen to the data.
On Fri, Dec 3, 2010 at 7:39 AM, Jörn Kottmann <ko...@gmail.com> wrote:
> On 12/3/10 1:29 PM, Thilo Goetz wrote:
>>
>> On 12/3/2010 13:14, Jörn Kottmann wrote:
>> [...]
>>>
>>> Having room for additional things like a corpus project could be nice,
>>> because people who are interested in the source code might not
>>> want to check out the corpus (if we ever get it created).
>>
>> So you can either make it convenient to check out
>> everything, or to check out individual stuff. When
>> I check out an open source project I tend to get
>> everything and then later figure out if there is
>> stuff in there that I don't need. Maybe most people
>> are different, no idea.
>
> Do you know if it would be possible to check in GPL training?
> We cannot distribute it, but we could distribute models built
> from it.
>
> There is a private training data repository, but it takes ages
> to check it out, so that might be something we should not
> stuff into the trunk which contains the source code.
>
> Jörn
>
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 1:29 PM, Thilo Goetz wrote:
> On 12/3/2010 13:14, Jörn Kottmann wrote:
> [...]
>>
>> Having room for additional things like a corpus project could be nice,
>> because people who are interested in the source code might not
>> want to check out the corpus (if we ever get it created).
>
> So you can either make it convenient to check out
> everything, or to check out individual stuff. When
> I check out an open source project I tend to get
> everything and then later figure out if there is
> stuff in there that I don't need. Maybe most people
> are different, no idea.
Do you know if it would be possible to check in GPL training?
We cannot distribute it, but we could distribute models built
from it.
There is a private training data repository, but it takes ages
to check it out, so that might be something we should not
stuff into the trunk which contains the source code.
Jörn
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 13:14, Jörn Kottmann wrote:
[...]
>
> Having room for additional things like a corpus project could be nice,
> because people who are interested in the source code might not
> want to check out the corpus (if we ever get it created).
So you can either make it convenient to check out
everything, or to check out individual stuff. When
I check out an open source project I tend to get
everything and then later figure out if there is
stuff in there that I don't need. Maybe most people
are different, no idea.
--Thilo
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 17:06, Benson Margulies wrote:
> As a bit of community advice, I'd advise an attitude of picking an
> initial tree structure that works, and then argue about re-arranging
> it later. I can see the distaste for multiple svn co's, though a
> 'view' directory with externals would cure that.
+1 to getting started with something and rearranging
it later if necessary. I think we're all eager to
start committing stuff, at least the website.
Subversion is very forgiving about changing the
repository layout later, it's not like we're
making a decision for all eternity.
--Thilo
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
As a bit of community advice, I'd advise an attitude of picking an
initial tree structure that works, and then argue about re-arranging
it later. I can see the distaste for multiple svn co's, though a
'view' directory with externals would cure that.
Re: OpenNLP Site
Posted by Grant Ingersoll <gs...@apache.org>.
Very, very barebones site is now live on http://incubator.apache.org/opennlp/
-Grant
On Dec 4, 2010, at 2:40 PM, Thilo Goetz wrote:
> On 12/4/2010 20:35, Jason Baldridge wrote:
>> Okay, I can be on board with this. Perhaps we just need to provide a "if you
>> only want maxent" instruction in the documentation. If it seems we need to
>> decouple the packages in the future, then we can do so.
>
> Absolutely. Also, at this point in the game, it's a bit
> early to worry about those details, imho. The svn layout
> is orthogonal to the release/distribution issue.
>
> We probably want to do a grand total release first of all,
> so we clear all the legal hurdles. If later on we find
> that we want to roll separate releases of individual
> packages, we can always do that. Somebody just has to
> step up and do the work :-)
>
> --Thilo
>
>>
>> Jason
>>
>> On Sat, Dec 4, 2010 at 3:40 AM, Jörn Kottmann<ko...@gmail.com> wrote:
>>
>>> On 12/3/10 11:00 PM, Jason Baldridge wrote:
>>>
>>>> I think it is more of a conceptual thing. You download a package like
>>>> maxent
>>>> so you can build models for your own prediction task, and it would be nice
>>>> to not have lots of extra documentation and such getting in the way of
>>>> understanding how to get up and running with the software, etc. I just
>>>> think
>>>> it makes sense as a separate release unit that the OpenNLP toolkit depends
>>>> on without swallowing it whole.
>>>>
>>>>
>>> I clearly disagree here. Having a separate maxent distribution maybe a
>>> little more convenient
>>> for maxent only users, but it has other disadvantages.
>>>
>>> To release the tools project we would have to release maxent always before
>>> which
>>> will add a two - three week delay for every tools project release which
>>> depends on the maxent change.
>>> Its also more work to release one more item.
>>> Its also more overhead from the project structure, we need a separate
>>> documentation project
>>> for maxent, etc. Do you also want a separate website for maxent ?
>>> The testing is also tangled, there are no good tests for maxent, expect the
>>> quite
>>> extensive regression test we do for the tools. In the end both projects
>>> must be
>>> ready for release, and maxent can not be really released separately.
>>> In practice I think it means that we have usually to release both at the
>>> same time,
>>> expect for smaller minor fixes.
>>>
>>> Jörn
>>>
>>>
>>
>>
>
--------------------------
Grant Ingersoll
http://www.lucidimagination.com
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/4/2010 20:35, Jason Baldridge wrote:
> Okay, I can be on board with this. Perhaps we just need to provide a "if you
> only want maxent" instruction in the documentation. If it seems we need to
> decouple the packages in the future, then we can do so.
Absolutely. Also, at this point in the game, it's a bit
early to worry about those details, imho. The svn layout
is orthogonal to the release/distribution issue.
We probably want to do a grand total release first of all,
so we clear all the legal hurdles. If later on we find
that we want to roll separate releases of individual
packages, we can always do that. Somebody just has to
step up and do the work :-)
--Thilo
>
> Jason
>
> On Sat, Dec 4, 2010 at 3:40 AM, Jörn Kottmann<ko...@gmail.com> wrote:
>
>> On 12/3/10 11:00 PM, Jason Baldridge wrote:
>>
>>> I think it is more of a conceptual thing. You download a package like
>>> maxent
>>> so you can build models for your own prediction task, and it would be nice
>>> to not have lots of extra documentation and such getting in the way of
>>> understanding how to get up and running with the software, etc. I just
>>> think
>>> it makes sense as a separate release unit that the OpenNLP toolkit depends
>>> on without swallowing it whole.
>>>
>>>
>> I clearly disagree here. Having a separate maxent distribution maybe a
>> little more convenient
>> for maxent only users, but it has other disadvantages.
>>
>> To release the tools project we would have to release maxent always before
>> which
>> will add a two - three week delay for every tools project release which
>> depends on the maxent change.
>> Its also more work to release one more item.
>> Its also more overhead from the project structure, we need a separate
>> documentation project
>> for maxent, etc. Do you also want a separate website for maxent ?
>> The testing is also tangled, there are no good tests for maxent, expect the
>> quite
>> extensive regression test we do for the tools. In the end both projects
>> must be
>> ready for release, and maxent can not be really released separately.
>> In practice I think it means that we have usually to release both at the
>> same time,
>> expect for smaller minor fixes.
>>
>> Jörn
>>
>>
>
>
Re: OpenNLP Site
Posted by Jason Baldridge <jb...@mail.utexas.edu>.
Okay, I can be on board with this. Perhaps we just need to provide a "if you
only want maxent" instruction in the documentation. If it seems we need to
decouple the packages in the future, then we can do so.
Jason
On Sat, Dec 4, 2010 at 3:40 AM, Jörn Kottmann <ko...@gmail.com> wrote:
> On 12/3/10 11:00 PM, Jason Baldridge wrote:
>
>> I think it is more of a conceptual thing. You download a package like
>> maxent
>> so you can build models for your own prediction task, and it would be nice
>> to not have lots of extra documentation and such getting in the way of
>> understanding how to get up and running with the software, etc. I just
>> think
>> it makes sense as a separate release unit that the OpenNLP toolkit depends
>> on without swallowing it whole.
>>
>>
> I clearly disagree here. Having a separate maxent distribution maybe a
> little more convenient
> for maxent only users, but it has other disadvantages.
>
> To release the tools project we would have to release maxent always before
> which
> will add a two - three week delay for every tools project release which
> depends on the maxent change.
> Its also more work to release one more item.
> Its also more overhead from the project structure, we need a separate
> documentation project
> for maxent, etc. Do you also want a separate website for maxent ?
> The testing is also tangled, there are no good tests for maxent, expect the
> quite
> extensive regression test we do for the tools. In the end both projects
> must be
> ready for release, and maxent can not be really released separately.
> In practice I think it means that we have usually to release both at the
> same time,
> expect for smaller minor fixes.
>
> Jörn
>
>
--
Jason Baldridge
Assistant Professor, Department of Linguistics
The University of Texas at Austin
http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by Grant Ingersoll <gs...@apache.org>.
On Dec 4, 2010, at 4:40 AM, Jörn Kottmann wrote:
> On 12/3/10 11:00 PM, Jason Baldridge wrote:
>> I think it is more of a conceptual thing. You download a package like maxent
>> so you can build models for your own prediction task, and it would be nice
>> to not have lots of extra documentation and such getting in the way of
>> understanding how to get up and running with the software, etc. I just think
>> it makes sense as a separate release unit that the OpenNLP toolkit depends
>> on without swallowing it whole.
>>
>
> I clearly disagree here. Having a separate maxent distribution maybe a little more convenient
> for maxent only users, but it has other disadvantages.
>
> To release the tools project we would have to release maxent always before which
> will add a two - three week delay for every tools project release which depends on the maxent change.
> Its also more work to release one more item.
> Its also more overhead from the project structure, we need a separate documentation project
> for maxent, etc. Do you also want a separate website for maxent ?
We had this problem in Lucene/Solr and eventually merged the two so that Solr could always be on the latest version of Lucene.
That being said, it doesn't necessarily imply a separate website, etc.
> The testing is also tangled, there are no good tests for maxent, expect the quite
> extensive regression test we do for the tools. In the end both projects must be
> ready for release, and maxent can not be really released separately.
> In practice I think it means that we have usually to release both at the same time,
> expect for smaller minor fixes.
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 11:00 PM, Jason Baldridge wrote:
> I think it is more of a conceptual thing. You download a package like maxent
> so you can build models for your own prediction task, and it would be nice
> to not have lots of extra documentation and such getting in the way of
> understanding how to get up and running with the software, etc. I just think
> it makes sense as a separate release unit that the OpenNLP toolkit depends
> on without swallowing it whole.
>
I clearly disagree here. Having a separate maxent distribution maybe a
little more convenient
for maxent only users, but it has other disadvantages.
To release the tools project we would have to release maxent always
before which
will add a two - three week delay for every tools project release which
depends on the maxent change.
Its also more work to release one more item.
Its also more overhead from the project structure, we need a separate
documentation project
for maxent, etc. Do you also want a separate website for maxent ?
The testing is also tangled, there are no good tests for maxent, expect
the quite
extensive regression test we do for the tools. In the end both projects
must be
ready for release, and maxent can not be really released separately.
In practice I think it means that we have usually to release both at the
same time,
expect for smaller minor fixes.
Jörn
Re: OpenNLP Site
Posted by Jason Baldridge <jb...@mail.utexas.edu>.
I think it is more of a conceptual thing. You download a package like maxent
so you can build models for your own prediction task, and it would be nice
to not have lots of extra documentation and such getting in the way of
understanding how to get up and running with the software, etc. I just think
it makes sense as a separate release unit that the OpenNLP toolkit depends
on without swallowing it whole.
On Fri, Dec 3, 2010 at 10:01 AM, Joern Kottmann <ko...@gmail.com> wrote:
> Isnt a binary didtribution always a download where you have to pull out a
> jar? A few kb more package size do not really matter IMHO. Jörn
> On Dec 3, 2010 4:50 PM, "Jason Baldridge" <jb...@mail.utexas.edu>
> wrote:
> > There are lots of people who use the maxent software for various
> prediction
> > tasks, and I think they would find it a pain to have to get the full
> > download and then pull out the jar. Basically, the maxent/ml stuff serves
> a
> > wider purpose than just the OpenNLP Toolkit, so I think it should have
> more
> > separation.
> >
> > On Fri, Dec 3, 2010 at 6:14 AM, Jörn Kottmann <ko...@gmail.com>
> wrote:
> >
> >> On 12/3/10 12:55 PM, Thilo Goetz wrote:
> >>
> >>> To my mind, it's totally a matter of taste. And I like
> >>> this better ;-) My main reason would be that it provides
> >>> a central trunk (which is what most everybody needs most
> >>> of the time). You can just check out everything with
> >>> svn co http://..../trunk opennlp, cd into opennlp and
> >>> build. You don't have to go to 3 different places to
> >>> get maxent, the toolkit and the UIMA wrappers.
> >>>
> >>
> >> It used to be that all three projects had to be checked
> >> out separately which I always found a little annoying.
> >>
> >> Also do not understand why we should not release
> >> all three together, users which just want
> >> maxent have the option to either just use maven (which some do)
> >> or copy over the maxent jar file from the OpenNLP distributable.
> >>
> >> Having room for additional things like a corpus project could be nice,
> >> because people who are interested in the source code might not
> >> want to check out the corpus (if we ever get it created).
> >>
> >> Jörn
> >>
> >
> >
> >
> > --
> > Jason Baldridge
> > Assistant Professor, Department of Linguistics
> > The University of Texas at Austin
> > http://comp.ling.utexas.edu/people/jason_baldridge
>
--
Jason Baldridge
Assistant Professor, Department of Linguistics
The University of Texas at Austin
http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by Joern Kottmann <ko...@gmail.com>.
Isnt a binary didtribution always a download where you have to pull out a
jar? A few kb more package size do not really matter IMHO. Jörn
On Dec 3, 2010 4:50 PM, "Jason Baldridge" <jb...@mail.utexas.edu> wrote:
> There are lots of people who use the maxent software for various
prediction
> tasks, and I think they would find it a pain to have to get the full
> download and then pull out the jar. Basically, the maxent/ml stuff serves
a
> wider purpose than just the OpenNLP Toolkit, so I think it should have
more
> separation.
>
> On Fri, Dec 3, 2010 at 6:14 AM, Jörn Kottmann <ko...@gmail.com> wrote:
>
>> On 12/3/10 12:55 PM, Thilo Goetz wrote:
>>
>>> To my mind, it's totally a matter of taste. And I like
>>> this better ;-) My main reason would be that it provides
>>> a central trunk (which is what most everybody needs most
>>> of the time). You can just check out everything with
>>> svn co http://..../trunk opennlp, cd into opennlp and
>>> build. You don't have to go to 3 different places to
>>> get maxent, the toolkit and the UIMA wrappers.
>>>
>>
>> It used to be that all three projects had to be checked
>> out separately which I always found a little annoying.
>>
>> Also do not understand why we should not release
>> all three together, users which just want
>> maxent have the option to either just use maven (which some do)
>> or copy over the maxent jar file from the OpenNLP distributable.
>>
>> Having room for additional things like a corpus project could be nice,
>> because people who are interested in the source code might not
>> want to check out the corpus (if we ever get it created).
>>
>> Jörn
>>
>
>
>
> --
> Jason Baldridge
> Assistant Professor, Department of Linguistics
> The University of Texas at Austin
> http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by Jason Baldridge <jb...@mail.utexas.edu>.
There are lots of people who use the maxent software for various prediction
tasks, and I think they would find it a pain to have to get the full
download and then pull out the jar. Basically, the maxent/ml stuff serves a
wider purpose than just the OpenNLP Toolkit, so I think it should have more
separation.
On Fri, Dec 3, 2010 at 6:14 AM, Jörn Kottmann <ko...@gmail.com> wrote:
> On 12/3/10 12:55 PM, Thilo Goetz wrote:
>
>> To my mind, it's totally a matter of taste. And I like
>> this better ;-) My main reason would be that it provides
>> a central trunk (which is what most everybody needs most
>> of the time). You can just check out everything with
>> svn co http://..../trunk opennlp, cd into opennlp and
>> build. You don't have to go to 3 different places to
>> get maxent, the toolkit and the UIMA wrappers.
>>
>
> It used to be that all three projects had to be checked
> out separately which I always found a little annoying.
>
> Also do not understand why we should not release
> all three together, users which just want
> maxent have the option to either just use maven (which some do)
> or copy over the maxent jar file from the OpenNLP distributable.
>
> Having room for additional things like a corpus project could be nice,
> because people who are interested in the source code might not
> want to check out the corpus (if we ever get it created).
>
> Jörn
>
--
Jason Baldridge
Assistant Professor, Department of Linguistics
The University of Texas at Austin
http://comp.ling.utexas.edu/people/jason_baldridge
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 13:26, Jörn Kottmann wrote:
> On 12/3/10 1:14 PM, Jörn Kottmann wrote:
>> On 12/3/10 12:55 PM, Thilo Goetz wrote:
>>> To my mind, it's totally a matter of taste. And I like
>>> this better ;-) My main reason would be that it provides
>>> a central trunk (which is what most everybody needs most
>>> of the time). You can just check out everything with
>>> svn co http://..../trunk opennlp, cd into opennlp and
>>> build. You don't have to go to 3 different places to
>>> get maxent, the toolkit and the UIMA wrappers.
>>
>> It used to be that all three projects had to be checked
>> out separately which I always found a little annoying.
>>
>
> I use to work a lot on the command line, having three
> different trunks mean, I have to go into each and type
> svn update three times in the end to get the latest code.
> Instead of typing it just once.
>
> Jörn
Yes exactly, couldn't agree more.
--Thilo
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 1:14 PM, Jörn Kottmann wrote:
> On 12/3/10 12:55 PM, Thilo Goetz wrote:
>> To my mind, it's totally a matter of taste. And I like
>> this better ;-) My main reason would be that it provides
>> a central trunk (which is what most everybody needs most
>> of the time). You can just check out everything with
>> svn co http://..../trunk opennlp, cd into opennlp and
>> build. You don't have to go to 3 different places to
>> get maxent, the toolkit and the UIMA wrappers.
>
> It used to be that all three projects had to be checked
> out separately which I always found a little annoying.
>
I use to work a lot on the command line, having three
different trunks mean, I have to go into each and type
svn update three times in the end to get the latest code.
Instead of typing it just once.
Jörn
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 12:55 PM, Thilo Goetz wrote:
> To my mind, it's totally a matter of taste. And I like
> this better ;-) My main reason would be that it provides
> a central trunk (which is what most everybody needs most
> of the time). You can just check out everything with
> svn co http://..../trunk opennlp, cd into opennlp and
> build. You don't have to go to 3 different places to
> get maxent, the toolkit and the UIMA wrappers.
It used to be that all three projects had to be checked
out separately which I always found a little annoying.
Also do not understand why we should not release
all three together, users which just want
maxent have the option to either just use maven (which some do)
or copy over the maxent jar file from the OpenNLP distributable.
Having room for additional things like a corpus project could be nice,
because people who are interested in the source code might not
want to check out the corpus (if we ever get it created).
Jörn
Re: OpenNLP Site
Posted by Thilo Goetz <tw...@gmx.de>.
On 12/3/2010 02:41, Benson Margulies wrote:
> Mine was a typo.
>
> You can always decide to release three things together. But if you
> lump them under one ttb, it's a pain to release them separately. So
Why?
> you *might*, and I really wish to defer to you guys, want to split
> them up upon import.
>
> What I really care about, however, is a structure where it is not too
> painful to change your/our minds later. So long as the overall tree
> has the extra level to allow for multiple TTB structures, it doesn't
> so much matter if you start with 1 or 4.
I don't buy into that argument. I see no problem with
trunk
proj1
proj2
proj3
branch
proj1
proj1branch1
...
proj2
tags
proj1
To my mind, it's totally a matter of taste. And I like
this better ;-) My main reason would be that it provides
a central trunk (which is what most everybody needs most
of the time). You can just check out everything with
svn co http://..../trunk opennlp, cd into opennlp and
build. You don't have to go to 3 different places to
get maxent, the toolkit and the UIMA wrappers.
BTW, we have the kind of setup that Benson suggests in
UIMA, and I find it annoying. So I speak from experience.
Of course, as I say above, it's pretty much a matter of
taste, and I'll go with the majority.
--Thilo
>
>
> On Thu, Dec 2, 2010 at 8:36 PM, Jörn Kottmann<ko...@gmail.com> wrote:
>> On 12/3/10 2:20 AM, James Kosin wrote:
>>>
>>> Thanks,
>>> Benson, that should be a /tags directory. /tag wouldn't work....
>>>
>>> Other than that it looks good.
>>>
>>> James
>>
>> /tag actual does work, since its really only a convention
>> (its usually a best practice to follow the convention).
>> Its even possible to use subversion without trunk, tags
>> and branch. Like we might do with the site folder.
>>
>> Jörn
>>
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
Mine was a typo.
You can always decide to release three things together. But if you
lump them under one ttb, it's a pain to release them separately. So
you *might*, and I really wish to defer to you guys, want to split
them up upon import.
What I really care about, however, is a structure where it is not too
painful to change your/our minds later. So long as the overall tree
has the extra level to allow for multiple TTB structures, it doesn't
so much matter if you start with 1 or 4.
On Thu, Dec 2, 2010 at 8:36 PM, Jörn Kottmann <ko...@gmail.com> wrote:
> On 12/3/10 2:20 AM, James Kosin wrote:
>>
>> Thanks,
>> Benson, that should be a /tags directory. /tag wouldn't work....
>>
>> Other than that it looks good.
>>
>> James
>
> /tag actual does work, since its really only a convention
> (its usually a best practice to follow the convention).
> Its even possible to use subversion without trunk, tags
> and branch. Like we might do with the site folder.
>
> Jörn
>
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 2:20 AM, James Kosin wrote:
> Thanks,
> Benson, that should be a /tags directory. /tag wouldn't work....
>
> Other than that it looks good.
>
> James
/tag actual does work, since its really only a convention
(its usually a best practice to follow the convention).
Its even possible to use subversion without trunk, tags
and branch. Like we might do with the site folder.
Jörn
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 8:07 PM, Jörn Kottmann wrote:
> On 12/3/10 1:57 AM, James Kosin wrote:
>> On 12/2/2010 7:35 PM, Benson Margulies wrote:
>>> James,
>>>
>>> It all depends on your proposed release architecture. Every releasable
>>> thing, in my opinion, gets its own TTB.
>>>
>>> That's the idea of the /products dir in the proposal I sent.
>>>
>>> --benson
>>>
>> Sorry, didn't get your proposed structure? Did I miss something?
>>
>
> The proposed structure from Benson can be found here:
> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3CAANLkTik3=kyvoOKpKueFkkakLVTPHS3cPq2g0W=dOD08@mail.gmail.com%3E
>
>
> Jörn
Thanks,
Benson, that should be a /tags directory. /tag wouldn't work....
Other than that it looks good.
James
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 1:57 AM, James Kosin wrote:
> On 12/2/2010 7:35 PM, Benson Margulies wrote:
>> James,
>>
>> It all depends on your proposed release architecture. Every releasable
>> thing, in my opinion, gets its own TTB.
>>
>> That's the idea of the /products dir in the proposal I sent.
>>
>> --benson
>>
> Sorry, didn't get your proposed structure? Did I miss something?
>
The proposed structure from Benson can be found here:
http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3CAANLkTik3=kyvoOKpKueFkkakLVTPHS3cPq2g0W=dOD08@mail.gmail.com%3E
Jörn
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 7:35 PM, Benson Margulies wrote:
> James,
>
> It all depends on your proposed release architecture. Every releasable
> thing, in my opinion, gets its own TTB.
>
> That's the idea of the /products dir in the proposal I sent.
>
> --benson
>
Sorry, didn't get your proposed structure? Did I miss something?
James
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
James,
It all depends on your proposed release architecture. Every releasable
thing, in my opinion, gets its own TTB.
That's the idea of the /products dir in the proposal I sent.
--benson
On Thu, Dec 2, 2010 at 7:13 PM, James Kosin <ja...@gmail.com> wrote:
> Hmmm...
>
> Maybe this would be better:
>
> /opennlp
> /tools
> /tags
> /branches
> /trunk
> /uima
> /tags
> /branches
> /trunk
> /maxent
> /tags
> /branches
> /trunk
>
> I know it looks odd, but, when we decide to move maxent on to the other
> project, It won't be as destructive and allows us to keep the three a
> bit separated.
> What is Apache's requirement on releases? Do they have to be a single
> tar? Or can it be several independent library modules?
>
> This way we can still have:
> opennlp-maxent-incubator-.......
> opennlp-uima-incubator-......
> opennlp-tools-incubator-......
>
> James
>
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
Hmmm...
Maybe this would be better:
/opennlp
/tools
/tags
/branches
/trunk
/uima
/tags
/branches
/trunk
/maxent
/tags
/branches
/trunk
I know it looks odd, but, when we decide to move maxent on to the other
project, It won't be as destructive and allows us to keep the three a
bit separated.
What is Apache's requirement on releases? Do they have to be a single
tar? Or can it be several independent library modules?
This way we can still have:
opennlp-maxent-incubator-.......
opennlp-uima-incubator-......
opennlp-tools-incubator-......
James
Re: OpenNLP Site
Posted by Benson Margulies <bi...@gmail.com>.
On Thu, Dec 2, 2010 at 6:18 PM, James Kosin <ja...@gmail.com> wrote:
> On 12/2/2010 6:10 PM, Jörn Kottmann wrote:
>> No that is the correct location:
>> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3C4CF81406.3060103@gmail.com%3E
>>
>>
>> Jörn
>>
>
> I thought svn took care of the branch and trunk directories. They are
> structural directories that get created when you commit. Then branches
> will contain all branches and the trunk gets updated. Tags also get
> their own directories.
Nope. Nothing is automatic. TTB is a convention, you have to create
those dirs in svn. If you release with maven, it will automatically
make tags, but nothing will make them.
>
> Just some of my school experiences with SVN.
>
> James
>
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 6:10 PM, Jörn Kottmann wrote:
> No that is the correct location:
> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3C4CF81406.3060103@gmail.com%3E
>
>
> Jörn
>
I thought svn took care of the branch and trunk directories. They are
structural directories that get created when you commit. Then branches
will contain all branches and the trunk gets updated. Tags also get
their own directories.
Just some of my school experiences with SVN.
James
Re: OpenNLP Site
Posted by James Kosin <ja...@gmail.com>.
On 12/2/2010 6:10 PM, Jörn Kottmann wrote:
> No that is the correct location:
> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3C4CF81406.3060103@gmail.com%3E
>
>
> Jörn
>
I guess everyone will have to do some reading...
http://svnbook.red-bean.com/
James
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
No that is the correct location:
http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/%3C4CF81406.3060103@gmail.com%3E
Jörn
On 12/3/10 12:09 AM, Jörn Kottmann wrote:
> On 12/3/10 12:05 AM, Jörn Kottmann wrote:
>>
>> But before we can do that we should get consensus on the project
>> subversion layout. So lets wait until tomorrow then everyone
>> else at least had a chance to see the proposed layout and express
>> their opinion.
>>
>
> That is the thread about the layout:
> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/browser
>
>
> Jörn
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/3/10 12:05 AM, Jörn Kottmann wrote:
>
> But before we can do that we should get consensus on the project
> subversion layout. So lets wait until tomorrow then everyone
> else at least had a chance to see the proposed layout and express
> their opinion.
>
That is the thread about the layout:
http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201012.mbox/browser
Jörn
Re: OpenNLP Site
Posted by Jörn Kottmann <ko...@gmail.com>.
On 12/2/10 11:59 PM, Grant Ingersoll wrote:
> I would suggest we setup and use the new Apache CMS for our site. See http://www.apache.org/dev/cms.html
>
> Thoughts? I can put in the request and get the initial stuff setup.
Yes there was already consensus to use the new CMS, I already said,
that I will create an issue over at the Infrastructure jira to set it up.
But before we can do that we should get consensus on the project
subversion layout. So lets wait until tomorrow then everyone
else at least had a chance to see the proposed layout and express
their opinion.
Jörn