You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@roller.apache.org by David M Johnson <Da...@Sun.COM> on 2006/01/25 17:41:09 UTC

Proposal Atlas: better support needs of large, diverse blog communities

I'm looking for feed back on Atlas, a new proposal for to extend  
Roller's built-in "planet" community aggregator so that it can better  
support the needs of large, diverse communities of bloggers.

http://rollerweblogger.org/wiki/Wiki.jsp?page=Proposal_Atlas

Here's the summary:

Currently Roller's front page displays an aggregation of the N most  
recent blog posts in every blog in the system and, on a separate tab,  
an aggregation of those same blogs plus a set of external blogs  
defined within Roller's planet aggregator. On large blog sites,  
that's just too much information. We need to make it easy to 1) find  
and follow (via syndication) blogs on specific topics of interest and  
in specific languages, 2) find the active blogs and topics of  
discussion and 3) find users and individual weblogs within a Roller  
system. This proposal describes how we can better support planet  
functionality in Roller (thus the name Atlas).

- Dave


Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Allen Gilliland <Al...@Sun.COM>.
On Thu, 2006-01-26 at 12:10, David M Johnson wrote:
> > In my opinion, the Planet Roller stuff is independent enough to  
> > warrant its own project separate from Roller.  Right now we don't  
> > really have that because there is planet code scattered throughout  
> > the Roller core codebase.  I think we should investigate the idea  
> > of managing the Planet stuff in a separate project that users can  
> > "add" to Roller via a few integration steps if desired.  Basically,  
> > decoupling the Planet code.
> 
> I don't want to do this now, but I will adjust the proposal so that  
> it does not preclude us from splitting off a pure planet app in the  
> future.
> 

agreed.  i don't think we need to split out the planet stuff immediately
either.  one thing that may be a nice middle ground would be to put the
planet code into it's own package, but still in the same source tree
that we have now.  something like org.planetroller.* or
org.roller.planet.*

-- Allen



Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Allen Gilliland <Al...@Sun.COM>.
On Thu, 2006-01-26 at 12:10, David M Johnson wrote:
> > In my opinion, the Planet Roller stuff is independent enough to  
> > warrant its own project separate from Roller.  Right now we don't  
> > really have that because there is planet code scattered throughout  
> > the Roller core codebase.  I think we should investigate the idea  
> > of managing the Planet stuff in a separate project that users can  
> > "add" to Roller via a few integration steps if desired.  Basically,  
> > decoupling the Planet code.
> 
> I don't want to do this now, but I will adjust the proposal so that  
> it does not preclude us from splitting off a pure planet app in the  
> future.
> 

agreed.  i don't think we need to split out the planet stuff immediately
either.  one thing that may be a nice middle ground would be to put the
planet code into it's own package, but still in the same source tree
that we have now.  something like org.planetroller.* or
org.roller.planet.*

-- Allen



Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Jean Vence <jv...@gmail.com>.
I think this is one of the most important aspect that needs to be
addressed.
Many people are integrating Roller with other applications. It would be
great if there was some mechanism that can be used by these applications to
create futher segmentation. You mention the ability to search by User,
categories, etc. How about making flexible enough for administrators to
create their own segmentation (they can chose to do it by department, team,
etc) by adding a mechanism that can be manipulated by other applications and
used to segment access and listing in Roller? This might be partially
achieved by manipulating the current categories feature but this approach is
not ideal because the user can always manipulate these categories. Thinking
out loud, perhaps you should implement site categories that cannot be
modified by a user and then within these categories, users may create their
own categories.

Just a thought.

Jean

On 1/26/06, David M Johnson <Da...@sun.com> wrote:
>
>
> On Jan 25, 2006, at 2:11 PM, Allen Gilliland wrote:
> > I'll take first stab.
> > This is a very big proposal so I think that some of these things
> > may be better off in their own proposals and with their own
> > discussions, so we aren't trying to talk about too many things at
> > once, but here goes.
>
> Thanks and yes, I may have to break this up as it grows.
>
>
> > 1. The proposal is centered very heavily around the idea to "Extend
> > Roller's built-in 'planet' community aggregator" and while I would
> > definitely like to see the planet stuff get more use I'm not sure
> > that we should be assuming that people want to use the planet
> > aggregator in their Roller install.
>
> Good point.
>
>
> > In my opinion, the Planet Roller stuff is independent enough to
> > warrant its own project separate from Roller.  Right now we don't
> > really have that because there is planet code scattered throughout
> > the Roller core codebase.  I think we should investigate the idea
> > of managing the Planet stuff in a separate project that users can
> > "add" to Roller via a few integration steps if desired.  Basically,
> > decoupling the Planet code.
>
> I don't want to do this now, but I will adjust the proposal so that
> it does not preclude us from splitting off a pure planet app in the
> future.
>
>
> > That being said, my main point is just that I don't think we should
> > be designing the new front page assuming that most people want to
> > use the planet stuff.  I think the front page should not contain
> > any idea about the planet stuff and can optionally be extended to
> > include planet features.
>
> Another good point. I'll change the proposal so that the front page
> does not rely directly on the planet tables or concepts.
>
>
> > 2. Assuming we stick with jsps and tiles for the frontpage, I'm not
> > convinced that the new jsp tags you suggested are the best way to
> > go.  Maybe I don't fully understand how things will work yet, but I
> > would prefer simple jstl for those functions rather than custom
> > tags.  My problem with custom tags is that they are hard to extend/
> > modify for users who want to do so.  Pure jsp code with jstl is
> > much easier to modify.
>
> The tags just give us a way to reuse those components on different
> JSP pages. I guess we could accomplish the same thing with Tiles.
>
>
> > 1. The proposal doesn't include any mention about the idea that
> > someone would want to configure a single blog as the front page.  I
> > believe there are a number of reasons to support this and I think
> > it should be part of this design plan.
>
> > 2. I think we should also consider the idea of managing the front
> > page via velocity templates which *could* be edited via the admin
> > UI.  It may turn out that this isn't the best way to go, but it's
> > worth considering.
>
> I'd like to work these ideas into the proposal as well.
>
> I'm starting to think that we should ditch the JSP front page
> entirely. Why shouldn't the front page be just as easily customized
> (via the Roller UI) as the blog pages themselves? There have been
> several times that we've wanted to change the front-pages at runtime
> and/or let an HTML/CSS designer work on the look-and-feel via the
> Roller UI.
>
> - Dave
>
>
>
>

Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by David M Johnson <Da...@Sun.COM>.
On Jan 25, 2006, at 2:11 PM, Allen Gilliland wrote:
> I'll take first stab.
> This is a very big proposal so I think that some of these things  
> may be better off in their own proposals and with their own  
> discussions, so we aren't trying to talk about too many things at  
> once, but here goes.

Thanks and yes, I may have to break this up as it grows.


> 1. The proposal is centered very heavily around the idea to "Extend  
> Roller's built-in 'planet' community aggregator" and while I would  
> definitely like to see the planet stuff get more use I'm not sure  
> that we should be assuming that people want to use the planet  
> aggregator in their Roller install.

Good point.


> In my opinion, the Planet Roller stuff is independent enough to  
> warrant its own project separate from Roller.  Right now we don't  
> really have that because there is planet code scattered throughout  
> the Roller core codebase.  I think we should investigate the idea  
> of managing the Planet stuff in a separate project that users can  
> "add" to Roller via a few integration steps if desired.  Basically,  
> decoupling the Planet code.

I don't want to do this now, but I will adjust the proposal so that  
it does not preclude us from splitting off a pure planet app in the  
future.


> That being said, my main point is just that I don't think we should  
> be designing the new front page assuming that most people want to  
> use the planet stuff.  I think the front page should not contain  
> any idea about the planet stuff and can optionally be extended to  
> include planet features.

Another good point. I'll change the proposal so that the front page  
does not rely directly on the planet tables or concepts.


> 2. Assuming we stick with jsps and tiles for the frontpage, I'm not  
> convinced that the new jsp tags you suggested are the best way to  
> go.  Maybe I don't fully understand how things will work yet, but I  
> would prefer simple jstl for those functions rather than custom  
> tags.  My problem with custom tags is that they are hard to extend/ 
> modify for users who want to do so.  Pure jsp code with jstl is  
> much easier to modify.

The tags just give us a way to reuse those components on different  
JSP pages. I guess we could accomplish the same thing with Tiles.


> 1. The proposal doesn't include any mention about the idea that  
> someone would want to configure a single blog as the front page.  I  
> believe there are a number of reasons to support this and I think  
> it should be part of this design plan.

> 2. I think we should also consider the idea of managing the front  
> page via velocity templates which *could* be edited via the admin  
> UI.  It may turn out that this isn't the best way to go, but it's  
> worth considering.

I'd like to work these ideas into the proposal as well.

I'm starting to think that we should ditch the JSP front page  
entirely. Why shouldn't the front page be just as easily customized  
(via the Roller UI) as the blog pages themselves? There have been  
several times that we've wanted to change the front-pages at runtime  
and/or let an HTML/CSS designer work on the look-and-feel via the  
Roller UI.

- Dave




Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by David M Johnson <Da...@Sun.COM>.
On Jan 25, 2006, at 2:11 PM, Allen Gilliland wrote:
> I'll take first stab.
> This is a very big proposal so I think that some of these things  
> may be better off in their own proposals and with their own  
> discussions, so we aren't trying to talk about too many things at  
> once, but here goes.

Thanks and yes, I may have to break this up as it grows.


> 1. The proposal is centered very heavily around the idea to "Extend  
> Roller's built-in 'planet' community aggregator" and while I would  
> definitely like to see the planet stuff get more use I'm not sure  
> that we should be assuming that people want to use the planet  
> aggregator in their Roller install.

Good point.


> In my opinion, the Planet Roller stuff is independent enough to  
> warrant its own project separate from Roller.  Right now we don't  
> really have that because there is planet code scattered throughout  
> the Roller core codebase.  I think we should investigate the idea  
> of managing the Planet stuff in a separate project that users can  
> "add" to Roller via a few integration steps if desired.  Basically,  
> decoupling the Planet code.

I don't want to do this now, but I will adjust the proposal so that  
it does not preclude us from splitting off a pure planet app in the  
future.


> That being said, my main point is just that I don't think we should  
> be designing the new front page assuming that most people want to  
> use the planet stuff.  I think the front page should not contain  
> any idea about the planet stuff and can optionally be extended to  
> include planet features.

Another good point. I'll change the proposal so that the front page  
does not rely directly on the planet tables or concepts.


> 2. Assuming we stick with jsps and tiles for the frontpage, I'm not  
> convinced that the new jsp tags you suggested are the best way to  
> go.  Maybe I don't fully understand how things will work yet, but I  
> would prefer simple jstl for those functions rather than custom  
> tags.  My problem with custom tags is that they are hard to extend/ 
> modify for users who want to do so.  Pure jsp code with jstl is  
> much easier to modify.

The tags just give us a way to reuse those components on different  
JSP pages. I guess we could accomplish the same thing with Tiles.


> 1. The proposal doesn't include any mention about the idea that  
> someone would want to configure a single blog as the front page.  I  
> believe there are a number of reasons to support this and I think  
> it should be part of this design plan.

> 2. I think we should also consider the idea of managing the front  
> page via velocity templates which *could* be edited via the admin  
> UI.  It may turn out that this isn't the best way to go, but it's  
> worth considering.

I'd like to work these ideas into the proposal as well.

I'm starting to think that we should ditch the JSP front page  
entirely. Why shouldn't the front page be just as easily customized  
(via the Roller UI) as the blog pages themselves? There have been  
several times that we've wanted to change the front-pages at runtime  
and/or let an HTML/CSS designer work on the look-and-feel via the  
Roller UI.

- Dave




Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Allen Gilliland <Al...@Sun.COM>.
I'll take first stab.

This is a very big proposal so I think that some of these things may be better off in their own proposals and with their own discussions, so we aren't trying to talk about too many things at once, but here goes ...

Some things I'm not sure about:

1. The proposal is centered very heavily around the idea to "Extend Roller's built-in 'planet' community aggregator" and while I would definitely like to see the planet stuff get more use I'm not sure that we should be assuming that people want to use the planet aggregator in their Roller install.

In my opinion, the Planet Roller stuff is independent enough to warrant its own project separate from Roller.  Right now we don't really have that because there is planet code scattered throughout the Roller core codebase.  I think we should investigate the idea of managing the Planet stuff in a separate project that users can "add" to Roller via a few integration steps if desired.  Basically, decoupling the Planet code.

That being said, my main point is just that I don't think we should be designing the new front page assuming that most people want to use the planet stuff.  I think the front page should not contain any idea about the planet stuff and can optionally be extended to include planet features.

2. Assuming we stick with jsps and tiles for the frontpage, I'm not convinced that the new jsp tags you suggested are the best way to go.  Maybe I don't fully understand how things will work yet, but I would prefer simple jstl for those functions rather than custom tags.  My problem with custom tags is that they are hard to extend/modify for users who want to do so.  Pure jsp code with jstl is much easier to modify.


Stuff that I really like:

1. The internal subscription idea.  I'd like more details on how the engineering behind that would work, but I think it's a very cool idea.

2. The browsable directory of profiles and weblogs.  Definitely something we should have.


Other ideas:

1. The proposal doesn't include any mention about the idea that someone would want to configure a single blog as the front page.  I believe there are a number of reasons to support this and I think it should be part of this design plan.

2. I think we should also consider the idea of managing the front page via velocity templates which *could* be edited via the admin UI.  It may turn out that this isn't the best way to go, but it's worth considering.

-- Allen


On Wed, 2006-01-25 at 08:41, David M Johnson wrote:
> I'm looking for feed back on Atlas, a new proposal for to extend  
> Roller's built-in "planet" community aggregator so that it can better  
> support the needs of large, diverse communities of bloggers.
> 
> http://rollerweblogger.org/wiki/Wiki.jsp?page=Proposal_Atlas



Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Miguel A Paraz <mp...@gmail.com>.
On 1/26/06, David M Johnson <Da...@sun.com> wrote:
> I'm looking for feed back on Atlas, a new proposal for to extend
> Roller's built-in "planet" community aggregator so that it can better
> support the needs of large, diverse communities of bloggers.

Nice. I'd like to help out.

My requested feature: I'd like sets of blogs ("planets?") to be owned
by individual users and not just the admin. If I sign up as a Roller
user, I'd get a blog plus my own planet page where I can choose what
blogs to include. If two or more users pick the same blog, the Planet
system only needs to fetch it once.

Re: Proposal Atlas: better support needs of large, diverse blog communities

Posted by Allen Gilliland <Al...@Sun.COM>.
I'll take first stab.

This is a very big proposal so I think that some of these things may be better off in their own proposals and with their own discussions, so we aren't trying to talk about too many things at once, but here goes ...

Some things I'm not sure about:

1. The proposal is centered very heavily around the idea to "Extend Roller's built-in 'planet' community aggregator" and while I would definitely like to see the planet stuff get more use I'm not sure that we should be assuming that people want to use the planet aggregator in their Roller install.

In my opinion, the Planet Roller stuff is independent enough to warrant its own project separate from Roller.  Right now we don't really have that because there is planet code scattered throughout the Roller core codebase.  I think we should investigate the idea of managing the Planet stuff in a separate project that users can "add" to Roller via a few integration steps if desired.  Basically, decoupling the Planet code.

That being said, my main point is just that I don't think we should be designing the new front page assuming that most people want to use the planet stuff.  I think the front page should not contain any idea about the planet stuff and can optionally be extended to include planet features.

2. Assuming we stick with jsps and tiles for the frontpage, I'm not convinced that the new jsp tags you suggested are the best way to go.  Maybe I don't fully understand how things will work yet, but I would prefer simple jstl for those functions rather than custom tags.  My problem with custom tags is that they are hard to extend/modify for users who want to do so.  Pure jsp code with jstl is much easier to modify.


Stuff that I really like:

1. The internal subscription idea.  I'd like more details on how the engineering behind that would work, but I think it's a very cool idea.

2. The browsable directory of profiles and weblogs.  Definitely something we should have.


Other ideas:

1. The proposal doesn't include any mention about the idea that someone would want to configure a single blog as the front page.  I believe there are a number of reasons to support this and I think it should be part of this design plan.

2. I think we should also consider the idea of managing the front page via velocity templates which *could* be edited via the admin UI.  It may turn out that this isn't the best way to go, but it's worth considering.

-- Allen


On Wed, 2006-01-25 at 08:41, David M Johnson wrote:
> I'm looking for feed back on Atlas, a new proposal for to extend  
> Roller's built-in "planet" community aggregator so that it can better  
> support the needs of large, diverse communities of bloggers.
> 
> http://rollerweblogger.org/wiki/Wiki.jsp?page=Proposal_Atlas