+[[!tag patch patch/core]]
+
I like the idea of [[tips/integrated_issue_tracking_with_ikiwiki]], and I do so on several wikis. However, as far as I can tell, ikiwiki has no functionality which can represent dependencies between bugs and allow pagespecs to select based on dependencies. For instance, I can't write a pagespec which selects all bugs with no dependencies on bugs not marked as done. --[[JoshTriplett]]
> I started having a think about this. I'm going to start with the idea that expanding
>>>> To fix that I'll need to pass a reference to that array into pagespec_makeperl.
>>>> I think I can then do the same thing to $params{specFuncs}. -- [[Will]]
+>>>>> You're right -- I did not think the recursive case through.
+>>>>> --[[Joey]]
+
> * Seems that the only reason `match_glob` has to check for `~` is
> because when a named spec appears in a pagespec, it is translated
> to `match_glob("~foo")`. If, instead, `pagespec_makeperl` checked
>>>> and
- define(aStar, a*) and link(aStar)
+ define(aStar, a*) and link(~aStar)
>>>> In the first case, we want the pagespec to match any page that links to a page matching the glob.
>>>> In the second case, we want the pagespec to match any page that links to a page matching the named spec.
>>>> match_link() was already doing existential part. The patches to this code were simply to remove the `lc()`
>>>> call from the named pagespec name. Can that `lc` be removed entirely? -- [[Will]]
+>>>>> I think we could get rid of it. `bestlink` will lc it itself
+>>>>> if the uppercase version does not exist; `match_glob` matches
+>>>>> insensitively.
+>>>>> --[[Joey]]
+
> * Generally, the need to modify `match_*` functions so that they
> check for and handle named pagespecs seems suboptimal, if
> only because there might be others people may want to use named
>>>> Yes, and it can do that in just three lines of code. But if we automatically check for named pagespecs all the time we
>>>> potentially break any matching function that doesn't accept pages, or wants to use multiple arguments.
+>>>>> 3 lines of code, plus the functions called become part of the API,
+>>>>> don't forget about that..
+>>>>>
+>>>>> Yes, I think that is the tradeoff, the question is whether to export
+>>>>> the additional complexity needed for that flexability.
+>>>>>
+>>>>> I'd be suprised if multiple argument pagespecs become necessary..
+>>>>> with the exception of this patch there has been no need for them yet.
+>>>>>
+>>>>> There are lots of pagespecs that take data other than pages,
+>>>>> indeed, that's really the common case. So far, none of them
+>>>>> seem likely to take data that starts with a `~`. Perhaps
+>>>>> the thing to do would be to check if `~foo` is a known,
+>>>>> named pagespec, and if not, just pass it through unchanged.
+>>>>> Then there's little room for ambiguity, and this also allows
+>>>>> pagespecs like `glob(~foo*)` to match the literal page `~foo`.
+>>>>> (It will make pagespec_merge even harder tho.. see below.)
+>>>>> --[[Joey]]
+
+>>>>>> I've already used multi-argument pagespec match functions in
+>>>>>> my data plugin. It is used for having different types of links. If
+>>>>>> you want to have multiple types of links, then the match function
+>>>>>> for them needs to take both the link name and the link type.
+>>>>>> I'm trying to think of a way we could have both - automatically
+>>>>>> handle the existential case unless the function indicates somehow
+>>>>>> that it'll do it itself. Any ideas? -- [[Will]]
+
> * I need to check if your trick to avoid infinite recursion
> works if there are two named specs that recursively
> call one-another. I suspect it does, but will test this
> --[[Joey]]
>> There is one issue that I've been thinking about that I haven't raised anywhere (or checked myself), and that is how this all interacts with page dependencies.
->> Firstly, I'm not sure anymore that the `pagespec_merge` function will continue to work in all cases.
+>>
+>>> I've moved the discussion of that to [[dependency_types]]. --[[Joey]]
+>>
+>> I'm not sure anymore that the `pagespec_merge` function will continue to work in all cases.
>>> The problem I can see there is that if two pagespecs
>>> get merged and both use `~foo` but define it differently,
>>> it shouldn't (but I haven't verified that really happens).
>>> That could certianly be a show-stopper. --[[Joey]]
+>>>> I think this can happen in the new closure based code. I don't think this could happen in the old code. -- [[Will]]
+
>>>> Even if that works, this is a good argument for having a syntactic difference between named pagespecs and normal pages.
>>>> If you're joining two pagespecs with 'or', you don't want a named pagespec in the first part overriding a page name in the
>>>> second part. Oh, and I assume 'or' has the right operator precedence that "a and b or c" is "(a and b) or c", and not "a and (b or c)" -- [[Will]]
>>>>> Looks like its bracketed in the code anyway... -- [[Will]]
->> Secondly, it seems that there are two types of dependency, and ikiwiki
->> currently only handles one of them. The first type is "Rebuild this
->> page when any of these other pages changes" - ikiwiki handles this.
->> The second type is "rebuild this page when set of pages referred to by
->> this pagespec changes" - ikiwiki doesn't seem to handle this. I
->> suspect that named pagespecs would make that second type of dependency
->> more important. I'll try to come up with a good example. -- [[Will]]
-
->>> Hrm, I was going to build an example of this with backlinks, but it
->>> looks like that is handled as a special case at the moment (line 458 of
->>> render.pm). I'll see if I can breapk
->>> things another way. Fixing this properly would allow removal of that special case. -- [[Will]]
-
->>>> I can't quite understand the distinction you're trying to draw
->>>> between the two types of dependencies. Backlinks are a very special
->>>> case though and I'll be suprised if they fit well into pagespecs.
->>>> --[[Joey]]
-
->>>>> The issue is that the existential pagespec matching allows you to build things that have similar
->>>>> problems to backlinks.
->>>>> e.g. the following inline:
-
- \[[!inline pages="define(~done, link(done)) and link(~done)" archive=yes]]
-
->>>>> includes any page that links to a page that links to done. Now imagine I add a new link to 'done' on
->>>>> some random page somewhere - a page which some other page links to which didn't previously get included - the set of pages accepted by the pagespec, and hence the set of
->>>>> pages inlined, will change. But, there is no dependency anywhere on the page that I altered, so
->>>>> ikiwiki will not rebuild the page with the inline in it. What is happening is that the page that I altered affects
->>>>> the set of pages matched by the pagespec without itself being matched by the pagespec, and hence included in the dependency list.
-
->>>>> To make this work well, I think you need to recognise two types of dependencies for each page (and no
->>>>> special cases for particular types of links, eg backlinks). The first type of dependency says, "The content of
->>>>> this page depends upon the content of these other pages". The `add_depends()` in the shortcuts
->>>>> plugin is of this form: any time the shortcuts page is edited, any page with a shortcut on it
->>>>> is rebuilt. The inline plugin also needs to add dependencies of this form to detect when the inlined
->>>>> content changes. By contrast, the map plugin does not need a dependency of this form, because it
->>>>> doesn't actually care about the content of any pages, just which pages it needs to include (which we'll handle next).
-
->>>>> The second type of dependency says, "The content of this page depends upon the exact set of pages matched
->>>>> by this pagespec". The first type of dependency was about the content of some pages, the second type is about
->>>>> which pages get matched by a pagespec. This is the type of dependency tracking that the map plugin needs.
->>>>> If the set of pages matched by map pagespec changes, then the page with the map on it needs to be rebuilt to show a different list of pages.
->>>>> Inline needs this type of dependency as well as the previous type - This type handles a change in which pages
->>>>> are inlined, the previous type handles a change in the content of any of those pages. Shortcut does not need this type of
->>>>> dependency. Most of the places that use `add_depends()` seem to need this type of dependency rather than the first type.
-
->>>>> Implementation Details: The first type of dependency can be handled very similarly to the current
->>>>> dependency system. You just need to keep a list of pages that the content depends upon. You could
->>>>> keep that list as a pagespec, but if you do this you might want to check that the pagespec doesn't change,
->>>>> possibly by adding a dependency of the second type along with the dependency of the first type.
-
->>>>> The second type of dependency is a little more tricky. For each page, we'd need a list of pagespecs that
->>>>> the page depended on, and for each pagespec you'd want to store the list of pages that currently match it.
->>>>> On refresh, you'd need to check each pagespec to see if the set of pages that match it has changed, and if
->>>>> that set has changed, then rebuild the dependent page(s). Oh, and for this second type of dependency, I
->>>>> don't think you can merge pagespecs. If I wanted to know if either "\*" or "link(done)" changes, then just checking
->>>>> to see if the set of pages matched by "\* or link(done)" changes doesn't work.
-
->>>>> The current system works because even though you usually want dependencies of the second type, the set of pages
->>>>> referred to by a pagespec can only change if one of those pages itself changes. i.e. A dependency check of the
->>>>> first type will catch a dependency change of the second type with current pagespecs.
->>>>> This doesn't work with backlinks, and it doesn't work with existential matching. Backlinks are currently special-cased. I don't know
->>>>> how to special-case existential matching - I suspect you're better off just getting the dependency tracking right.
-
->>>>> I also tried to come up with other possible solutions: e.g. can we find the dependencies for a pagespec? That
->>>>> would be the set of pages where a change on one of those pages could lead to a change in the set of pages matched by the pagespec.
->>>>> For old-style pagespecs without backlinks, the dependency set for a pagespec is the same as the set of pages the pagespec matches.
->>>>> Unfortunately, with existential matching, the set of pages that each
->>>>> pagespec depends upon can quickly become "*", which is not very useful. -- [[Will]]
+>>>> Perhaps the thing to do is to have a `clear_defines()`
+>>>> function, then merging `A` and `B` yields `(A) or (clear_defines() and (B))`
+>>>> That would deal with both the cases where `A` and `B` differently
+>>>> define `~foo` as well as with the case where `A` defines `~foo` while
+>>>> `B` uses it to refer to a literal page.
+>>>> --[[Joey]]
+
+>>>>> I don't think this will work with the new patch, and I don't think it was needed with the old one.
+>>>>> Under the old patch, pagespec_makeperl() generated a string of unevaluated, self-contained, perl
+>>>>> code. When a new named pagespec was defined, a recursive call was made to get the perl code
+>>>>> for the pagespec, and then that code was used to add something like `$params{specFuncs}->{name} = sub {recursive code} and `
+>>>>> to the result of the calling function. This means that at pagespec testing time, when this code is executed, the
+>>>>> specFuncs hash is built up as the pagespec is checked. In the case of the 'or' used above, later redefinitions of
+>>>>> a named pagespec would have redefined the specFunc at the right time. It should have just worked. However...
+
+>>>>> Since my original patch, you started using closures for security reasons (and I can see the case for that). Unfortunately this
+>>>>> means that the generated perl code is no longer self-contained - it needs to be evaluated in the same closure it was generated
+>>>>> so that it has access to the data array. To make this work with the recursive call I had two options: a) make the data array a
+>>>>> reference that I pass around through the pagespec_makeperl() functions and have available when the code is finally evaluated
+>>>>> in pagespec_translate(), or b) make sure that each pagespec is evaluated in its correct closure and a perl function is returned, not a
+>>>>> string containing unevaluated perl code.
+
+>>>>> I went with option b). I did it in such a way that the hash of specfuncs is built up at translation time, not at execution time. This
+>>>>> means that with the new code you can call specfuncs that get defined out of order:
+
+ ~test and define(~test, blah)
+
+>>>>> but it also means that using a simple 'or' to join two pagespecs wont work. If you do something like this:
+
+ ~test and define(~test, foo) and define(~test, baz)
+
+>>>>> then the last definition (baz) takes precedence.
+>>>>> In the process of writing this I think I've come up with a way to change this back the way it was, still using closures. -- [[Will]]
+
+>>> My [[remove-pagespec-merge|should_optimise_pagespecs]] branch has now
+>>> solved all this by deleting the offending function :-) --[[smcv]]
+
+
Patch updated to use closures rather than inline generated code for named pagespecs. Also includes some new use of ErrorReason where appropriate. -- [[Will]]
+> * Perl really doesn't need forward declarations, honest!
+
+>> It complained (warning, not error) when I didn't use the forward declaration. :(
+
+> * I have doubts about memoizing the anonymous sub created by
+> `pagespec_translate`.
+
+>> This is there explicitly to make sure that runtime is polynomial and not exponential.
+
+> * Think where you wrote `+{}` you can just write `{}`
+
+>> Possibly :) -- [[Will]]
+
----
diff --git a/IkiWiki.pm b/IkiWiki.pm