You mentioned you fear Component gets too big, I think the API slowly
gets bloated and this an opportunity to get rid of 2 methods which are
IMO not needed. The number of methods has grown over time, that is
natural and nothing to worry about, but from time to time one should
re-think implementation, pattern etc and apply lessons learned and if
there is an opportunity to improve in 2.0, than we should do it.
Juergen
On 10/2/06, Matej Knopp <[EMAIL PROTECTED]> wrote:
I'm not I understand this. If this worked as you suggest, then
component.markupId may not be transient, right? Otherwise you'd have to
call getMarkupId not only in constructor, but every time the page is
deserialized.
What would be the benefits of removing setOutputMarkupId and
getOutputMarkupId()?
-Matej
Juergen Donnerstag wrote:
> Actually we could get rid of Component.FLAG_OUTPUT_MARKUP_ID,
> Component.getOutputMarkupId() and Component.setOutputMarkupId() as
> well.
>
> If Component.getMarkupId() were implemented like
> public String getMarkupId()
> {
> String id = getMarkupAttributes().getString("id");
> if (id == null)
> {
> if (this.markupId != null)
> {
> return this.markupId;
> }
>
> id = getPageRelativePath();
> // first escape _ with __
> id = id.replace("_", "__");
> // then replace : with _
> id = id.replace(':', '_');
>
> // cache the value
> this.markupId = id;
> }
> return id;
> }
>
> than you only need to call getMarkupId() once to make sure the id is
> properly set and rendered in the output.
> May getMarkupId() is not the right the name than, what about
> enableMarkupId() or requiresMarkupId(boolean)
>
> Juergen
>
> On 10/2/06, Juergen Donnerstag <[EMAIL PROTECTED]> wrote:
>> The size of Component is not different than in 1.2, as far as I'm
>> aware nothing has changed so far except that Johan added
>> markupAttributes. Replacing markupAttribute with a transient String
>> doesn't increase the size either. A transient would allow us cache the
>> id on first request (lazy) and re-compute in case of cluster / session
>> reload scenarios.
>>
>> Any yes, getMarkupAttributes() would be a read-only map.
>>
>> Juergen
>>
>> On 10/2/06, Matej Knopp <[EMAIL PROTECTED]> wrote:
>> > Yeah, but we don't need to store the id in component. It can be
>> computed
>> > when necessary. So even if we decide to cache it, we can clean the
>> > cached value on detach.
>> >
>> > I know it's just one string, but I'm starting to be concerned with the
>> > size of Component in 2.0.
>> >
>> > -Matej
>> >
>> > Juergen Donnerstag wrote:
>> > > See Component.java. It is required for JavaScripts
>> > >
>> > > /**
>> > > * Retrieves id by which this component is represented within the
>> > > markup.
>> > > * <p>
>> > > * If the id attribute is present in the markup attributes of
>> this
>> > > component
>> > > * it will be used, otherwise the page-relative path of this
>> > > component will
>> > > * be used.
>> > > *
>> > > * @return the Markup id
>> > > */
>> > > public String getMarkupId()
>> > > {
>> > > String id = getMarkupAttributes().getString("id");
>> > > if (id == null)
>> > > {
>> > > id = getPageRelativePath();
>> > > // first escape _ with __
>> > > id = id.replace("_", "__");
>> > > // then replace : with _
>> > > id = id.replace(':', '_');
>> > > getMarkupAttributes().put("id", id);
>> > > }
>> > > return id;
>> > > }
>> > >
>> > >
>> > > On 10/2/06, Matej Knopp <[EMAIL PROTECTED]> wrote:
>> > >> +1 to remove it. I never liked this copy on write value map. It's
>> > >> nothing I can't do in onComponentTag or using a behavior.
>> > >>
>> > >> I think method to retrieve tag attributes from markup would be
>> > >> sufficient.
>> > >>
>> > >> Btw. what would be the purpose of id variable?
>> > >>
>> > >> -Matej
>> > >>
>> > >> Juergen Donnerstag wrote:
>> > >> > Currently Component.java maintains a copy-on-write ValueMap for
>> markup
>> > >> > attributes which user might want to change, such as getMarkupId()
>> > >> > which adds the id="xxx" attribute to the tag.
>> > >> >
>> > >> > However, the current implementation doesn't work properly as the
>> > >> > Component owned attribute map keep the "old" value even when the
>> > >> > markup has been re-loaded because of locale, style or variation
>> > >> > changes (see FormExample)
>> > >> >
>> > >> > I wonder if this general use case of being able to modify any tag
>> > >> > attribute in the constructor realy exists or if "id" is the
>> only case.
>> > >> > If the latter is true, than I would like to suggest to remove
>> > >> > markupAttribute from Component as it requires some realy
>> hackish code
>> > >> > later on. I'd much rather add a "id" variable to the component
>> for the
>> > >> > specific purpose of MarkupId()..
>> > >> >
>> > >> > What do think?
>> > >> >
>> > >> > Juergen
>> > >> >
>> > >>
>> > >>
>> > >
>> >
>> >
>>
>
