[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
RE: [cdt-dev] Error Parser Manager removes leading whitespace
|
> -----Original Message-----
> From: cdt-dev-bounces@xxxxxxxxxxx
> [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Volanakis, Elias
> Sent: Wednesday, November 23, 2005 2:33 PM
> To: CDT General developers list.
> Subject: AW: [cdt-dev] Error Parser Manager removes leading whitespace
>
> Hi Pete,
>
> Pete MacLiesh wrote:
> >
> > It would actually be nice to eat continuations forwards
> > rather than getting them one at a time. We have a tool that
> > pumps out dissertations like
> >
> > Warning (TIE_WARN_UNUSED) foo.tie:27
> > the dissertation starts
> > here and goes on for
> > several lines making the
> > previous line alone not useful
> >
> >
> > So if an error parser could peek forwards and optionally eat
> > continuations that would make it more robust. Probably not
> > easy the way it is stuctured as a stream though.
>
> sorry if I misunderstood your posting, but the above seems a bit of a
> digression :).
Maybe a different way to solve the same problem (which would serve my needs better)
>
> I don't have an issue with receiving a "line at a time" from the Error
> Parser Manager.
Ah ... but I do. Peeking ahead (and eating continuations) would reduce instability caused
by error parsers like the LD one that match against almost anything. The parser sequencing
has been made deterministic in cdt 3.0, but is static so you can't dynamically stop
another error parser that "normally" precedes you. I have made maybe 10 different fixes in
error parsers over the last 2 years to prevent false matches (maybe specific to our tool
variants / versions (mostly gnu) but still irritating).
PMac
> My issue with the Error Parser Manager is,
> that it trims
> the leading spaces of each line of the tool output _before_ they are
> passed to our own Error Parser.
>
> The continuation detection problem can be solved nicely,
> without peeking
> beyond the current line, by making the Error Parser stateful
> (which some
> might say is the same as "peeking backwards" ;). This however is not
> something that affects the Error Parser Manager in any way.
>
> Just my 2 cents worth of opinion :).
>
> Kind regards,
> Elias.
> _______________________________________________
> cdt-dev mailing list
> cdt-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/cdt-dev
>
