Re: Typecache does not contain include tags created at expansion

Le 03/01/2015 21:20, Eric Ludlam a écrit :
> Using the 'dump' feature should force the include stream to be filled, 
> so there is some other reason why it is empty.  Since 'dependents' has 
> an entry, I'll assume that is your include tag.   It is possible that 
> since you are developing your grammar as you go, some buffers are in 
> different parsing states.  I recently patched CEDET in bzr (and now 
> git) back in.. November? to better update buffers when you compile 
> your grammar.
Thanks for your answer Eric. The entry in dependents is a scope cache 
containing the include tag and other thing. Here is the result of the 
typecache dump a bit more unfolded :

    ]#<semanticdb-typecache /<project
    dir>/src/AnotherNs/SuperGreeterCaller.php>
        ] Name: "/<project dir>/src/AnotherNs/SuperGreeterCaller.php"
        ] Class: #'semanticdb-typecache
        ] filestream #<TAG LIST: 1 entries>
          * AnotherNs : namespace
        ] includestream : nil
        ] stream : nil
        ] dependants #<list o' stuff: 1 entries>
          > #<semantic-scope-cache Cache>
              ] Name: "Cache"
              ] Class: #'semantic-scope-cache
              ] :table #<semanticdb-table SuperGreeterCaller.php (1 tags)>
              ] tag +callSuperGreeter
              ] scopetypes : nil
              ] parents #<TAG LIST: 2 entries>
              ] parentinheritance : nil
              ] scope #<TAG LIST: 4 entries>
              ] fullscope #<TAG LIST: 6 entries>
                * +callSuperGreeter
                * SuperGreeterCaller : class
                * /<project dir>/src/NewNs/SuperGreeter.php
                * Greeter : use
                * AnotherNs : namespace
                * SuperGreeterCaller : class
              ] localargs #<TAG LIST: 1 entries>
              ] localvar #<TAG LIST: 4 entries>
              ] typescope : nil

If you need it, you can reproduce my problem by installing php-mode 
(easy through MELPA), and then use these commands to checkout my code 
and run my ERT tests with "emacs -q" :

    git clone https://bitbucket.org/stevenremot/cedet
    cd cedet
    git checkout php-tests
    make
    cd contrib
    make
    cd tests/php
    sh run-test.sh

  2 tests should fail, "wisent-php-argument-completion-through-alias", 
and "wisent-php-in-depth-completion". If you then go to the buffer 
"SuperGreeterCaller.php", you can get the same result than me with 
"semanticdb-typecache-dump".

>
> A way to check is to get the above output, immediately visit the 
> included file, and use 'bovinate' in there to see what the content is. 
> If it has no types, it won't be in the typecache.  If it does, and you 
> go back and re-request the typecache and the content is there, then 
> there is some update issue.  Those are puzzling to debug.
I checked, and bovinate returns the expected types.
>
> In your output, I notice that you tags are saved in :members in an 
> order different than the text of the parsed buffer.  Is that a cut and 
> paste error in your email, or is that real?  I also noticed the output 
> excludes an positional information, either in [ start end ] or overlay 
> form.  Is that on purpose?
I think the reverse order in :members is not done on purpose. Would it 
be better to put them on the right order ? Also, I stripped position 
information to ease reading in the mail, but if you need it I can give 
you the full result of bovinate.
>
>
> Lastly, if "use" acts both as include and using statement, you could 
> try can use wisent-php-expand-tag to convert the include to also be a 
> 'using' statement at the same location.  That way the scope 
> calculation engine could identify and convert "Greeter" into the other 
> type when found.   I'm guessing a bit about that, so you could just 
> wait until you need it to give it a try.
In the PHP semantics, "use" behaves only as a using statement, but as 
there is no explicit "include" in modern PHP, using it as an include 
statement too seems to be the best thing I can do to include the aliased 
type in the scope.  I currently generate a compound "alias" tag in the 
grammar (because you can write "use A as B, C as D;"), and then for each 
aliasing I generate a type tag and an include tag as shown in my last 
mail. It is the result of experimentation of various methods until I 
find one that works perfectly, so if there a more elegant way do this 
I'm totally willing to use it.

I can't find anything about the "using" tag in the manual. I already saw 
it in the code for C++ support, is this tag specific to this language ?

Thank you again for your help.

------------------------------------------------------------------------------
Dive into the World of Parallel Programming! The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net

_______________________________________________
Cedet-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/cedet-devel