| Using an external compiler instead of java parser [message #29432] |
Mon, 20 July 2009 06:53  |
Eclipse User |
|
|
|
Originally posted by: erich.blanch1111.hotmail.com
Hi,
I have read that IMP can be used with other parser technologies than LPG.
But what if fhe "other" parser is an external compiler?
The background to my question is that we need to sort out (once for all) if
it is possible to re-use the front-end part of the existing compiler to
avoid the redunant/duplicate effort in writing almost the same
functionallity in Java. I have serous doubts about hooking in an external
compiler
during each reconcile event (to provide online error annotaions - which
implies that the source file must be saved prior to invoking the compiler)
and if an 3rd party compiler actually can provide such verbose output that
it makes scense.
Perhaps I'm missing something here and perhaps this is a commonly used
technology, can anyone give their opinion about this issue and perhaps point
to some project that actaully does this. I have found an Erlang IDE that
seems to be doing something like this but they are using
some background daemon server process that provides the analysis services.
Any help in this appreciated
regards
Erich
|
|
|
| Re: Using an external compiler instead of java parser [message #29505 is a reply to message #29432] |
Tue, 21 July 2009 01:43  |
| Eclipse User |
|
|
|
Hi Erich
I think that CDT now creates regular saves in order to exploit gcc. It
doesn't work very well because the line number context from gcc makes it
difficult to distinguish definition from reference so an error may
appear in many places; in practice the console has to be used to view
the gcc error log. That said, it is a considerable improvement over the
save, view, locate, edit cycle.
I don't think that CDT uses IMP or LPG.
Regards
Ed Willink
Erich wrote:
> Hi,
>
> I have read that IMP can be used with other parser technologies than LPG.
> But what if fhe "other" parser is an external compiler?
>
> The background to my question is that we need to sort out (once for all) if
> it is possible to re-use the front-end part of the existing compiler to
> avoid the redunant/duplicate effort in writing almost the same
> functionallity in Java. I have serous doubts about hooking in an external
> compiler
> during each reconcile event (to provide online error annotaions - which
> implies that the source file must be saved prior to invoking the compiler)
> and if an 3rd party compiler actually can provide such verbose output that
> it makes scense.
>
> Perhaps I'm missing something here and perhaps this is a commonly used
> technology, can anyone give their opinion about this issue and perhaps point
> to some project that actaully does this. I have found an Erlang IDE that
> seems to be doing something like this but they are using
> some background daemon server process that provides the analysis services.
>
> Any help in this appreciated
>
> regards
>
> Erich
>
>
>
>
>
>
|
|
|
| Re: Using an external compiler instead of java parser [message #576207 is a reply to message #29432] |
Tue, 21 July 2009 01:43 |
| Eclipse User |
|
|
|
Hi Erich
I think that CDT now creates regular saves in order to exploit gcc. It
doesn't work very well because the line number context from gcc makes it
difficult to distinguish definition from reference so an error may
appear in many places; in practice the console has to be used to view
the gcc error log. That said, it is a considerable improvement over the
save, view, locate, edit cycle.
I don't think that CDT uses IMP or LPG.
Regards
Ed Willink
Erich wrote:
> Hi,
>
> I have read that IMP can be used with other parser technologies than LPG.
> But what if fhe "other" parser is an external compiler?
>
> The background to my question is that we need to sort out (once for all) if
> it is possible to re-use the front-end part of the existing compiler to
> avoid the redunant/duplicate effort in writing almost the same
> functionallity in Java. I have serous doubts about hooking in an external
> compiler
> during each reconcile event (to provide online error annotaions - which
> implies that the source file must be saved prior to invoking the compiler)
> and if an 3rd party compiler actually can provide such verbose output that
> it makes scense.
>
> Perhaps I'm missing something here and perhaps this is a commonly used
> technology, can anyone give their opinion about this issue and perhaps point
> to some project that actaully does this. I have found an Erlang IDE that
> seems to be doing something like this but they are using
> some background daemon server process that provides the analysis services.
>
> Any help in this appreciated
>
> regards
>
> Erich
>
>
>
>
>
>
|
|
|
Powered by
FUDForum. Page generated in 0.03997 seconds
Home