Thanks Dirk, I wasn't aware of the findFreeEntry method. It basically
confirms that there is no built-in method to search for pages by pattern,
and that a linear search must be employed. Practically, I think checking
the first 10 or so would be sufficient, as it would be highly unusual to
write more than 10 blog entries on any given day, and then to delete most
of them.
I'll see what I can come up with.
Ulf
On Sat, Jun 22, 2024 at 10:35 PM Dirk Frederickx <dirk.frederi...@gmail.com>
wrote:
> Ulf,
>
> The problem seems to be in o.a.w.tags/CalenderTag.java in getDayLink().
>
> It only checks for the page with the pattern
> ${param.page}_blogentry_'ddMMyy'_1
> ; which has the suffix "_1".
>
> I guess it should actually also check for the existence of pages with a
> higher _nn.
> It should stop at MAX_BLOG_ENTRIES (now set tot 10_000), but that would be
> performance wise not realistic.
> Better would be to check for the existence of pages starting with the
> pattern ${param.page}_blogentry_'ddMMyy'_ followed by a /\d+/.
> Possibly you can find inspiration in o.a.w.plugin/WeblogEntryPlugin in
> findFreeEntry() which is used to create new entries.
>
>
> Grtz
> dirk
>
> On Fri, Jun 21, 2024 at 10:15 AM Ulf Dittmer
> <ulf.ditt...@googlemail.com.invalid> wrote:
>
> > The CalendarTag, when deciding whether to link a particular day, assumes
> > that the presence or absence of a page "..._blogentry_'ddMMyy'_1'
> indicates
> > whether there are weblog entries for a given day (this is in
> Sidebar.jsp).
> >
> > The problem is, if the first entry on that day (the one with ..._1) is
> > deleted, others on that day (..._2 etc.) are not checked, and so that day
> > is not linked.
> >
> > You can see this on
> > https://jspwiki-wiki.apache.org/Wiki.jsp?page=Haddock%20Weblog%20Example
> > I created two blog entries, and then deleted the first one. Today's date
> > (June 21) should still be highlighted and linked because of the second
> > entry, but it isn't.
> >
> > I'm not asking for this to be fixed, I'm just looking for advice on what
> > the best approach would be. If there's a reasonable solution, I'd be
> happy
> > to work on that and contribute it.
> >
> > Ulf
> >
>
