bug#703: Octave comments in emacs

bug#703: Octave comments in emacs

[Chong Yidong]

Hi Kurt,

Could you take a look at this bug report?

http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=697
http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=703

Thanks.

2008/8/11 Rosen Diankov <rdiankov@cs.cmu.edu>:

> The % also messes up formatting when seen in strings, ie
> 'my value: %d'
>
> For some reason, the octave syntax does not recognize that single
> quotes are strings (it might be because the transpose character is
> also ' and it is hard to differentiate between a string and the
> transposing).
>
> In any case, writing the above example will treat the % as a comment.
> The indenting engine will skip the closing parenthesis )... resulting
> in all the rest of the indents looking like garbage.
>
....
>
> In calculate-octave-indent, just delete the entire expression
>
> ((looking-at "\\s<\\S<")
>        (setq icol (list comment-column icol)))
>
> that forces comments to comment-column (which is 32). Was there a
> reason for this?

bug#703: Octave comments in emacs

[Stefan Monnier]

> Is there a way to assign two different syntax rules for the same
> character in emacs? Ie, one where ' is treated as a string quote (and
> suppresses the syntax elements inside the quotes), and one where it is
> treated as a suffix operator.

Yes, there is, using the `syntax-table' property.
A common way to set this property is via the
font-lock-syntactic-keywords variable.

So all it takes is a regexp good enough to distinguish the two cases ;-)


        Stefan

bug#703: Octave comments in emacs

[Kurt Hornik]

>>>>> Chong Yidong writes:

> Hi Kurt,
> Could you take a look at this bug report?

> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=697
> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=703

> Thanks.

Sure.

I am not quite sure I understand what the issue is.

Octave mode has never provided syntax support for single quoted strings
because there is no way (short of actually parsing the code) of telling
these apart from single quotes used for transposition.

The comment style provided by Octave mode is the same as for Lisp mode,
so a single comment char is aligned at the comment column.

Best
-k

> 2008/8/11 Rosen Diankov <rdiankov@cs.cmu.edu>:

>> The % also messes up formatting when seen in strings, ie
>> 'my value: %d'
>> 
>> For some reason, the octave syntax does not recognize that single
>> quotes are strings (it might be because the transpose character is
>> also ' and it is hard to differentiate between a string and the
>> transposing).
>> 
>> In any case, writing the above example will treat the % as a comment.
>> The indenting engine will skip the closing parenthesis )... resulting
>> in all the rest of the indents looking like garbage.
>> 
> ....
>> 
>> In calculate-octave-indent, just delete the entire expression
>> 
>> ((looking-at "\\s<\\S<")
>> (setq icol (list comment-column icol)))
>> 
>> that forces comments to comment-column (which is 32). Was there a
>> reason for this?

bug#703: Octave comments in emacs

[Rosen Diankov]

Hi all,

Coming from Matlab, I feel much more comfortable if the single quotes
were treated like double quotes (transposing is not as important), and
if the comments were aligned to the start of the next line (like the
// in C++). Given that a lot of people use Octave as a free
alternative to Matlab, I think a lot would expect the same formatting
rules as the matlab editor.

At least it would be great to expose these options to the user like
'octave-auto-indent' instead of having people hunt through
octave-mod.el to get the behavior they want.

Rosen,

2008/8/20 Kurt Hornik <Kurt.Hornik@wu-wien.ac.at>:
>>>>>> Chong Yidong writes:
>
>> Hi Kurt,
>> Could you take a look at this bug report?
>
>> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=697
>> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=703
>
>> Thanks.
>
> Sure.
>
> I am not quite sure I understand what the issue is.
>
> Octave mode has never provided syntax support for single quoted strings
> because there is no way (short of actually parsing the code) of telling
> these apart from single quotes used for transposition.
>
> The comment style provided by Octave mode is the same as for Lisp mode,
> so a single comment char is aligned at the comment column.
>
> Best
> -k
>
>> 2008/8/11 Rosen Diankov <rdiankov@cs.cmu.edu>:
>
>>> The % also messes up formatting when seen in strings, ie
>>> 'my value: %d'
>>>
>>> For some reason, the octave syntax does not recognize that single
>>> quotes are strings (it might be because the transpose character is
>>> also ' and it is hard to differentiate between a string and the
>>> transposing).
>>>
>>> In any case, writing the above example will treat the % as a comment.
>>> The indenting engine will skip the closing parenthesis )... resulting
>>> in all the rest of the indents looking like garbage.
>>>
>> ....
>>>
>>> In calculate-octave-indent, just delete the entire expression
>>>
>>> ((looking-at "\\s<\\S<")
>>> (setq icol (list comment-column icol)))
>>>
>>> that forces comments to comment-column (which is 32). Was there a
>>> reason for this?
>
>

bug#703: Octave comments in emacs

[Rosen Diankov]

And one other question, what would it take to get the Abstract Syntax
Tree for an Octave expression through elisp? Like Kurt said, we'll
need to parse the code in order to actually differentiate between the
transposing operator ' and a string. Isn't there anything in elisp
that will allow you to do this?

I'm willing to put some time into this if someone can start me off in
the right direction.

rosen,

2008/8/20 Rosen Diankov <rdiankov@cs.cmu.edu>:
> Hi all,
>
> Coming from Matlab, I feel much more comfortable if the single quotes
> were treated like double quotes (transposing is not as important), and
> if the comments were aligned to the start of the next line (like the
> // in C++). Given that a lot of people use Octave as a free
> alternative to Matlab, I think a lot would expect the same formatting
> rules as the matlab editor.
>
> At least it would be great to expose these options to the user like
> 'octave-auto-indent' instead of having people hunt through
> octave-mod.el to get the behavior they want.
>
> Rosen,
>
> 2008/8/20 Kurt Hornik <Kurt.Hornik@wu-wien.ac.at>:
>>>>>>> Chong Yidong writes:
>>
>>> Hi Kurt,
>>> Could you take a look at this bug report?
>>
>>> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=697
>>> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=703
>>
>>> Thanks.
>>
>> Sure.
>>
>> I am not quite sure I understand what the issue is.
>>
>> Octave mode has never provided syntax support for single quoted strings
>> because there is no way (short of actually parsing the code) of telling
>> these apart from single quotes used for transposition.
>>
>> The comment style provided by Octave mode is the same as for Lisp mode,
>> so a single comment char is aligned at the comment column.
>>
>> Best
>> -k
>>
>>> 2008/8/11 Rosen Diankov <rdiankov@cs.cmu.edu>:
>>
>>>> The % also messes up formatting when seen in strings, ie
>>>> 'my value: %d'
>>>>
>>>> For some reason, the octave syntax does not recognize that single
>>>> quotes are strings (it might be because the transpose character is
>>>> also ' and it is hard to differentiate between a string and the
>>>> transposing).
>>>>
>>>> In any case, writing the above example will treat the % as a comment.
>>>> The indenting engine will skip the closing parenthesis )... resulting
>>>> in all the rest of the indents looking like garbage.
>>>>
>>> ....
>>>>
>>>> In calculate-octave-indent, just delete the entire expression
>>>>
>>>> ((looking-at "\\s<\\S<")
>>>> (setq icol (list comment-column icol)))
>>>>
>>>> that forces comments to comment-column (which is 32). Was there a
>>>> reason for this?
>>
>>
>

bug#703: Octave comments in emacs

[Chong Yidong]

Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> > Is there a way to assign two different syntax rules for the same
> > character in emacs? Ie, one where ' is treated as a string quote
> > (and suppresses the syntax elements inside the quotes), and one
> > where it is treated as a suffix operator.
>
> Yes, there is, using the `syntax-table' property.  A common way to set
> this property is via the font-lock-syntactic-keywords variable.
>
> So all it takes is a regexp good enough to distinguish the two cases

Kurt Hornik <Kurt.Hornik@wu-wien.ac.at> wrote:

> Octave mode has never provided syntax support for single quoted
> strings because there is no way (short of actually parsing the code)
> of telling these apart from single quotes used for transposition.

Maybe we could use a rough heuristic to handle most of the cases.  A '
character that occurs before "[=(] *" is probably the beginning of a
single quoted string, not a transpose operator.  WDYT?

bug#703: Octave comments in emacs

[Rosen Diankov]

> Maybe we could use a rough heuristic to handle most of the cases.  A '
> character that occurs before "[=(] *" is probably the beginning of a
> single quoted string, not a transpose operator.  WDYT?
I can think of a lot of cases where this would break.

The proper way of doing it is to tokenize a line and emulate a simple
parser. Although the rules that will be used to parse octave
expressions don't need to be thourough, they have to be sufficient
enough to detect transposition vs strings.

As far as I know, context free grammars are strictly more powerful
than regular expressions. A simple example is keeping track of [] and
(). So it might be impossible to generate a regular expression doing
what we want it to do.

Environments like Visual Studio and Eclipse do dynamic code analysis
all the time, so I would be surprised if emacs doesn't have some
equivalent parsing functionality.

Rosen,

2008/8/21 Chong Yidong <cyd@stupidchicken.com>:
> Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
>> > Is there a way to assign two different syntax rules for the same
>> > character in emacs? Ie, one where ' is treated as a string quote
>> > (and suppresses the syntax elements inside the quotes), and one
>> > where it is treated as a suffix operator.
>>
>> Yes, there is, using the `syntax-table' property.  A common way to set
>> this property is via the font-lock-syntactic-keywords variable.
>>
>> So all it takes is a regexp good enough to distinguish the two cases
>
> Kurt Hornik <Kurt.Hornik@wu-wien.ac.at> wrote:
>
>> Octave mode has never provided syntax support for single quoted
>> strings because there is no way (short of actually parsing the code)
>> of telling these apart from single quotes used for transposition.
>
> Maybe we could use a rough heuristic to handle most of the cases.  A '
> character that occurs before "[=(] *" is probably the beginning of a
> single quoted string, not a transpose operator.  WDYT?
>