You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@wicket.apache.org by Susan Liebeskind <su...@gtri.gatech.edu> on 2008/11/03 15:31:16 UTC
Incremental migration from JSP-based applications to Wicket?
Hi,
This helpful posting
http://herebebeasties.com/2007-03-01/jsp-and-wicket-sitting-in-a-tree
is often cited as a guide whenever anyone asks about migrating from a
JSP-based web application to an equivalent JSP-free Wicket application.
But...
the posting itself is >18 months old and describes work with a
pre-release version of 1.3. Wicket 1.3.5 is out as of Oct 2008, with
1.4-M4 just around the corner. I also see that the posting clearly
states that this is not a boxed solution, or particularly
well-supported, buyer beware and all that. Fair enough. So it is with
low expectations, but with hope for an answer nonetheless :-), that I
ask the question
"is there a newer/better solution available now for supporting JSPs in
Wicket 1.3 or the upcoming Wicket 1.4, one which is
cleaner/easier/addresses gotchas mentioned here. Or is the technique
referenced here still best practice?"
My guess is that Alistair's technique is still best practice, based on
newer postings I see on this list, but I'd like to ask the question
outright, just in case I've missed something.
************************************************
Some thoughts about migrating a JSP based application to Wicket:
I know that in the long run, one would choose Wicket so as to be JSP
free. And certainly there are issues with JSP development as we all
know. I would almost say that to some developers, JSPs are the high
fructose corn syrup of the web development world, very sweet (in some
ways), but bad for you in the long haul :-)
Thing is, going JSP free is a transition if you have a legacy app.
There's where you are now, there's where you want to be, and a migration
path which involves some period of time where one would need to support
both JSP and Wicket pages in the same app.
In the case of our current project, we have some large Struts 1.x based
applications with many JSPs (nearly 900._ Some JSPs are simple, just
divs filled in by Ajax components, but other JSPs are complex and use
scriptlets, inline javascript, 3rd party and custom tags. There's more
work in recoding these JSPS (extracting the HTML template skeleton from
the JSPs, recoding the core/3rd party/custom tags from taglibs in Java,
replacing our simple use of Tiles 1 for page composition again with
Java, refactoring those scriptlets, calling out to a custom 3rd party
Ajax toolkit...etc. etc.).
The point is, we cannot simply stop, migrate the whole app over to
Wicket and bring it back up - a gradual migration is necessary, doing
JSP migration in bitesize chunks.
My colleague Rich Allen has been posting various questions to the this
list as we evaluate our replacement for Struts, and after looking at
Wicket some, I understand why he likes Wicket a lot. It is a clean
division of application from presentation, and for a new development
effort, that would be really cool. But we have a large legacy
application to support - that's just the way it is. So a
straightforward and incremental migration path from existing JSPs to
Wicket is something that would be critical for our team.
Costs to consider: We have a lot of developers, some experience and some
newbies, and we have a large code base. Replacing JSPs is not an
inexpensive proposition, and we need to do it in small pieces, and
clearly spell out the approach for our team. Indeed one argument for
choosing something like Spring MVC to replace Struts is that Spring MVC
is another MVC framework that, for better or worse (and I know the
community on this list would say "for worse" :-) , preserve our
investment in JSPs.
Again, if this were a new development, starting with a clean slate, the
decision inputs are different. But when you have substantial investment
in JSPs, a valid question is "do we want to pay the cost of retooling
now vs. later or not at all? Can we simply be disciplined in our use of
JSPs, but continue to use them nonetheless, because the cost of
retooling is too high?"
By choosing Wicket, we would be saying "enough with JSPs", but we have
to have a cost effective way of incrementally achieving that goal, where
cost is measured in how easy it is to retrain our developers and to
migrate in small, manageable pieces.
Any advice/anecdotes are gratefully welcomed.
Thanks in advance.
Susan
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by Martin Grigorov <mc...@e-card.bg>.
Hi Susan,
I'm in the same situation. With the only difference I have few months
advance.
And I'd say the project works very good for now. Currently some of the
pages are backed by Wicket and the rest are still JSP ones and there is
no difference from a customer point of view.
None of the Wicket pages does include any JSP chunks. They are made
entirely in the Wicket way. So we don't use AlMaw's solution at all. At
the beginning it was slow but now we started to reuse more and more
components and it's getting faster to migrate the next JSP page.
good luck
martin-g
On Mon, 2008-11-03 at 09:31 -0500, Susan Liebeskind wrote:
> Hi,
>
> This helpful posting
>
> http://herebebeasties.com/2007-03-01/jsp-and-wicket-sitting-in-a-tree
>
> is often cited as a guide whenever anyone asks about migrating from a
> JSP-based web application to an equivalent JSP-free Wicket application.
>
> But...
>
> the posting itself is >18 months old and describes work with a
> pre-release version of 1.3. Wicket 1.3.5 is out as of Oct 2008, with
> 1.4-M4 just around the corner. I also see that the posting clearly
> states that this is not a boxed solution, or particularly
> well-supported, buyer beware and all that. Fair enough. So it is with
> low expectations, but with hope for an answer nonetheless :-), that I
> ask the question
>
> "is there a newer/better solution available now for supporting JSPs in
> Wicket 1.3 or the upcoming Wicket 1.4, one which is
> cleaner/easier/addresses gotchas mentioned here. Or is the technique
> referenced here still best practice?"
>
> My guess is that Alistair's technique is still best practice, based on
> newer postings I see on this list, but I'd like to ask the question
> outright, just in case I've missed something.
>
> ************************************************
>
> Some thoughts about migrating a JSP based application to Wicket:
>
> I know that in the long run, one would choose Wicket so as to be JSP
> free. And certainly there are issues with JSP development as we all
> know. I would almost say that to some developers, JSPs are the high
> fructose corn syrup of the web development world, very sweet (in some
> ways), but bad for you in the long haul :-)
>
> Thing is, going JSP free is a transition if you have a legacy app.
> There's where you are now, there's where you want to be, and a migration
> path which involves some period of time where one would need to support
> both JSP and Wicket pages in the same app.
>
> In the case of our current project, we have some large Struts 1.x based
> applications with many JSPs (nearly 900._ Some JSPs are simple, just
> divs filled in by Ajax components, but other JSPs are complex and use
> scriptlets, inline javascript, 3rd party and custom tags. There's more
> work in recoding these JSPS (extracting the HTML template skeleton from
> the JSPs, recoding the core/3rd party/custom tags from taglibs in Java,
> replacing our simple use of Tiles 1 for page composition again with
> Java, refactoring those scriptlets, calling out to a custom 3rd party
> Ajax toolkit...etc. etc.).
>
> The point is, we cannot simply stop, migrate the whole app over to
> Wicket and bring it back up - a gradual migration is necessary, doing
> JSP migration in bitesize chunks.
>
> My colleague Rich Allen has been posting various questions to the this
> list as we evaluate our replacement for Struts, and after looking at
> Wicket some, I understand why he likes Wicket a lot. It is a clean
> division of application from presentation, and for a new development
> effort, that would be really cool. But we have a large legacy
> application to support - that's just the way it is. So a
> straightforward and incremental migration path from existing JSPs to
> Wicket is something that would be critical for our team.
>
> Costs to consider: We have a lot of developers, some experience and some
> newbies, and we have a large code base. Replacing JSPs is not an
> inexpensive proposition, and we need to do it in small pieces, and
> clearly spell out the approach for our team. Indeed one argument for
> choosing something like Spring MVC to replace Struts is that Spring MVC
> is another MVC framework that, for better or worse (and I know the
> community on this list would say "for worse" :-) , preserve our
> investment in JSPs.
>
> Again, if this were a new development, starting with a clean slate, the
> decision inputs are different. But when you have substantial investment
> in JSPs, a valid question is "do we want to pay the cost of retooling
> now vs. later or not at all? Can we simply be disciplined in our use of
> JSPs, but continue to use them nonetheless, because the cost of
> retooling is too high?"
>
> By choosing Wicket, we would be saying "enough with JSPs", but we have
> to have a cost effective way of incrementally achieving that goal, where
> cost is measured in how easy it is to retrain our developers and to
> migrate in small, manageable pieces.
>
> Any advice/anecdotes are gratefully welcomed.
>
> Thanks in advance.
>
> Susan
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by John Krasnay <jo...@krasnay.ca>.
On Wed, Nov 05, 2008 at 11:49:02AM -0500, Susan Liebeskind wrote:
> Igor Vaynberg wrote:
> >i think the best way to migrate an app is to migrate it a
> >page/pageflow at a time. running wicket inside a jsp or jsp inside
> >wicket is always going to have gotchas you will have to work around.
> >
> >
> Agreed, and you just have to live with some quirks/gotchas while you go
> through the migration.
>
> But I'm just looking for a sense of how people start actually using
> Wicket. Do you have any feel for the ratio of new development in Wicket
> vs. migration old apps to Wicket, just based on the traffic you've seen
> go by on the list over the years?
>
> Again, I know this is a non-scientific survey and not something that can
> be really quantified.
>
> Susan
I migrated an application from SpringMVC/Freemarker (no JSP) to Wicket
over a period of several weeks. I basically did it as Igor suggested, a
page[flow] at a time; some URLs went to the old pages, some to the new.
The biggest issue I can recall is with feedback panels. If we raised a
feedback message (e.g. Session.get().info("Record saved.")) then
redirected to a non-Wicket page, the user wouldn't see the message until
they later hit a Wicket page. I'm sure we could have worked something up
to display Wicket feedback messages from non-Wicket pages but we
completed the migration before it became much of an issue.
jk
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by Igor Vaynberg <ig...@gmail.com>.
it is difficult to even guess because even if some people are
migrating their apps they dont necessarily mention it. start a new
thread on the list, call it [survey] how many people are migrating
existing apps to wicket vs start new apps and see how people respond.
you also have to define migrate. some people might think that
rewriting a new version of an app from scratch is migrating also.
-igor
On Wed, Nov 5, 2008 at 8:49 AM, Susan Liebeskind
<su...@gtri.gatech.edu> wrote:
> Igor Vaynberg wrote:
>>
>> i think the best way to migrate an app is to migrate it a
>> page/pageflow at a time. running wicket inside a jsp or jsp inside
>> wicket is always going to have gotchas you will have to work around.
>>
>>
>
> Agreed, and you just have to live with some quirks/gotchas while you go
> through the migration.
> But I'm just looking for a sense of how people start actually using Wicket.
> Do you have any feel for the ratio of new development in Wicket vs.
> migration old apps to Wicket, just based on the traffic you've seen go by on
> the list over the years?
>
> Again, I know this is a non-scientific survey and not something that can be
> really quantified.
>
> Susan
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by Susan Liebeskind <su...@gtri.gatech.edu>.
Igor Vaynberg wrote:
> i think the best way to migrate an app is to migrate it a
> page/pageflow at a time. running wicket inside a jsp or jsp inside
> wicket is always going to have gotchas you will have to work around.
>
>
Agreed, and you just have to live with some quirks/gotchas while you go
through the migration.
But I'm just looking for a sense of how people start actually using
Wicket. Do you have any feel for the ratio of new development in Wicket
vs. migration old apps to Wicket, just based on the traffic you've seen
go by on the list over the years?
Again, I know this is a non-scientific survey and not something that can
be really quantified.
Susan
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by Igor Vaynberg <ig...@gmail.com>.
i think the best way to migrate an app is to migrate it a
page/pageflow at a time. running wicket inside a jsp or jsp inside
wicket is always going to have gotchas you will have to work around.
-igor
On Wed, Nov 5, 2008 at 7:47 AM, Susan Liebeskind
<su...@gtri.gatech.edu> wrote:
> A follow up to this question, which generated just a single response from
> another fellow traveler...
>
> The lack of response to questions about migrating JSP app could also be
> indicative of the fact that Wicket is being used mostly to develop *new*
> applications. In other words, maybe only a smallish percentage of folks are
> trying to migrate their existing work. Instead, maybe a more typical use
> case is that people begin new developments with Wicket, and don't try to
> port over existing one.
>
> Would some of the Wicket committers on this list have any anecdotal sense
> about whether Wicket is being used primarily for new apps vs, being a
> migration target?
>
> Thanks again.
>
> Susan
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Incremental migration from JSP-based applications to Wicket?
Posted by Susan Liebeskind <su...@gtri.gatech.edu>.
A follow up to this question, which generated just a single response
from another fellow traveler...
The lack of response to questions about migrating JSP app could also be
indicative of the fact that Wicket is being used mostly to develop
*new* applications. In other words, maybe only a smallish percentage of
folks are trying to migrate their existing work. Instead, maybe a more
typical use case is that people begin new developments with Wicket, and
don't try to port over existing one.
Would some of the Wicket committers on this list have any anecdotal
sense about whether Wicket is being used primarily for new apps vs,
being a migration target?
Thanks again.
Susan
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org