Looks like I'll use 9 new styles to do basic variable interpolation highlighting of $<identifier> and @<identifier> in the various quote[-like] operator strings:
"" qq{} `` qx{} // m{} qr{} s{}{} <<HERE
For things like qq{}, interpolation highlighting is not done when the a single quote delimiter is used. Skipping highlighting of more complex forms like $a[1], $b->{key} etc. Too much of talk of 'heuristics' in the documentation... leaving that for another day. Also skipping highlighting of interpolated special variables except for the usual $_ and @_.
This is roughly the behaviour of the oldish gvim and Komodo I have currently, although they do not seem to do variable interpolation highlighting for <<HERE blocks.
For the new style numbers, the plan is they are offset in value from the parent style by a fixed, identical constant, in case I want to use that behaviour. If this assumption is not desirable, let me know and I'll skip the use of this assumption.
-- Cheers, Kein-Hong Man (esq.) Kuala Lumpur, Malaysia
> Looks like I'll use 9 new styles to do basic variable
> interpolation highlighting of $<identifier> and @<identifier> in
> the various quote[-like] operator strings:
> Preliminary work code and test cases (simple variable interpolation > for double-quoted strings only):
VarInterpolation doesn't need to be marked with SCI_METHOD since it is only used locally. SCI_METHOD is there to force a particular calling convention (stdcall) so that lexers can be used if they are separately compiled from Scintilla with a different calling convention as may be the case with external lexers. Forcing stdcall on a method may stop compiler optimizations like using fastcall or thiscall.
> KHMan: >> Preliminary work code and test cases (simple variable interpolation >> for double-quoted strings only):
> VarInterpolation doesn't need to be marked with SCI_METHOD since it > is only used locally. SCI_METHOD is there to force a particular > calling convention (stdcall) so that lexers can be used if they are > separately compiled from Scintilla with a different calling convention > as may be the case with external lexers. Forcing stdcall on a method > may stop compiler optimizations like using fastcall or thiscall.
OK
> Otherwise, it looks good.
-- Cheers, Kein-Hong Man (esq.) Kuala Lumpur, Malaysia
> > KHMan:
> >> Preliminary work code and test cases (simple variable interpolation
> >> for double-quoted strings only):
> > VarInterpolation doesn't need to be marked with SCI_METHOD since it
> > is only used locally. SCI_METHOD is there to force a particular
> > calling convention (stdcall) so that lexers can be used if they are
> > separately compiled from Scintilla with a different calling convention
> > as may be the case with external lexers. Forcing stdcall on a method
> > may stop compiler optimizations like using fastcall or thiscall.
> OK
> > Otherwise, it looks good.
> --
> Cheers,
> Kein-Hong Man (esq.)
> Kuala Lumpur, Malaysia
> On Aug 12, 3:44 am, KHMan wrote: >> On 8/12/2011 7:26 AM, Neil Hodgson wrote:
>>> KHMan: >>>> Preliminary work code and test cases (simple variable interpolation >>>> for double-quoted strings only):
>>> VarInterpolation doesn't need to be marked with SCI_METHOD since it >>> is only used locally. SCI_METHOD is there to force a particular >>> calling convention (stdcall) so that lexers can be used if they are >>> separately compiled from Scintilla with a different calling convention >>> as may be the case with external lexers. Forcing stdcall on a method >>> may stop compiler optimizations like using fastcall or thiscall.
>> OK
>>> Otherwise, it looks good.
-- Cheers, Kein-Hong Man (esq.) Kuala Lumpur, Malaysia
I released it in as a *developer* version to CPAN as part of the
transition from Scintilla 1.x to scintilla latest. So it is not a big
deal. I am also working on a Perl 6 lexer for Scintilla. Basically
we're backporting Wx::STC binding to use scintilla latest over older
wxWidgets 2.8.x platforms.
Please keep us up-to-date with LexPerl.cxx changes. As soon as I have
a working LexPerl6.cxx, i'll post it on this group.
--
Thanks,
Ahmad M. Zawawi
On Aug 12, 2:25 pm, KHMan <keinh...@gmail.com> wrote:
> > Now Padre (http://padre.perlide.org) and other Wx-based Perl editors
> > can utilize your awesome patch right away.
> Hey, not so fast. It's just a preview, I pinged and got the
> feedback I need from Neil.
> I've done "" qq{} `` qx{}, hope to finish it before the end of the
> month. A few hundred more test cases to check...
> The final to-be-applied version will be posted to the Scintilla SF
> feature tracker. You can grab it once it gets into hg.
> > Note: 'perl -c perl-test-interpolation.pl' fails. Is it that intended?
> Hmmm... perhaps you should look at the contents of the file. :-)
> > On Aug 12, 3:44 am, KHMan wrote:
> >> On 8/12/2011 7:26 AM, Neil Hodgson wrote:
> >>> KHMan:
> >>>> Preliminary work code and test cases (simple variable interpolation
> >>>> for double-quoted strings only):
> >>> VarInterpolation doesn't need to be marked with SCI_METHOD since it
> >>> is only used locally. SCI_METHOD is there to force a particular
> >>> calling convention (stdcall) so that lexers can be used if they are
> >>> separately compiled from Scintilla with a different calling convention
> >>> as may be the case with external lexers. Forcing stdcall on a method
> >>> may stop compiler optimizations like using fastcall or thiscall.
> >> OK
> >>> Otherwise, it looks good.
> --
> Cheers,
> Kein-Hong Man (esq.)
> Kuala Lumpur, Malaysia
> I released it in as a *developer* version to CPAN as part of the > transition from Scintilla 1.x to scintilla latest. So it is not a big > deal. I am also working on a Perl 6 lexer for Scintilla. Basically > we're backporting Wx::STC binding to use scintilla latest over older > wxWidgets 2.8.x platforms.
> Please keep us up-to-date with LexPerl.cxx changes.
Variable interpolation is the last biggie that's missing from LexPerl. I don't remember anything else that's really worth doing for LexPerl...
> As soon as I have > a working LexPerl6.cxx, i'll post it on this group.
Good to know, and looking forward to writing some Perl 6 on SciTE. I never got started on doing anything serious on Perl 6. If you have a good set of tests for Perl 6 highlighting, I'd be interested to dabble in it in the (very vague) future. IMHO, it was possible to get Perl 5 highlighted mostly correctly mainly by amassing a large set of test cases.
-- Cheers, Kein-Hong Man (esq.) Kuala Lumpur, Malaysia
One thing to note, $self->SetStyleBits(7) is needed otherwise we'll have indicators all over the place. Am i doing something wrong here? GetStyleBitsNeeded() still reports only 5 needed bits.
> -- > Cheers, > Kein-Hong Man (esq.) > Kuala Lumpur, Malaysia
> -- > You received this message because you are subscribed to the Google Groups > "scintilla-interest" group. > To post to this group, send email to scintilla-interest@googlegroups.com. > To unsubscribe from this group, send email to > scintilla-interest+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/scintilla-interest?hl=en.
> One thing to note, $self->SetStyleBits(7) is needed otherwise we'll > have indicators all over the place. Am i doing something wrong here? > GetStyleBitsNeeded() still reports only 5 needed bits.
Yes, I think. (I hack lexers only, I'm all at sea with the other bits of Scintilla.)
-- Cheers, Kein-Hong Man (esq.) Kuala Lumpur, Malaysia
> One thing to note, $self->SetStyleBits(7) is needed otherwise we'll > have indicators all over the place. Am i doing something wrong here? > GetStyleBitsNeeded() still reports only 5 needed bits.
SCI_GETSTYLEBITSNEEDED returns 8 for me. The number of bits is set as the last parameter in the LexerModule constructor: LexerModule lmPerl(SCLEX_PERL, LexerPerl::LexerFactoryPerl, "perl", perlWordListDesc, 8);
> On 8/12/2011 9:14 PM, Ahmad M. Zawawi wrote:
> >[snip]
> > Please keep us up-to-date with LexPerl.cxx changes. As soon as I have
> > a working LexPerl6.cxx, i'll post it on this group.
Interim code. One or two more things to do before submitting to the
tracker.
One style added for tr{}{} y{}{} translations which do not have
interpolation.
Some colours adjusted for the SciTE properties file, but then I not
real picky. The adjustment is intended that any interpolated variable
in various types of quoted blocks is darkish-red. Users welcome to
propose changes if their eyes hurt... :-)
TODO:
(1) Special variables (other than $_ and @_ which already work). Will
need it otherwise users will be wondering why they do not work, except
for // m{} qr{} s{}{} (unless some kind of reliable ruleset can be
divined).
(2) ${bareword} @{bareword} if it does not involve ripping up too much
code. Elsewhere, things like "$a[$i]" looks quite nice already;
parsing the entire var spec seems like overkill.
--
Cheers,
Kein-Hong Man (esq.)
Kuala Lumpur, Malaysia
The kinds of additional patterns recognized are:
${word} @{word} and things like $${word}
$#{word} @#{word} and things like $#${word}
${digit} @{digit}
$$ and most special variables with operator chars
$0 to $9 and things like $123 (perl seem to greedily lex)
$^A and other control char special variables
InterpolateSegment() is the new code for variable interpolation. Once
the others (// m{} qr{} s{}{} here-docs) are shifted too, the older
VarInterpolation() will be deleted. The new code is closer to the way
Perl parses quoted operators and is less likely to exhibit glitches.
--
Cheers,
Kein-Hong Man (esq.)
Kuala Lumpur, Malaysia
