Eclipse Community Forums: TMF (Xtext)

Help

Home

Home » Modeling » TMF (Xtext) » Column based grammar

Show: Today's Messages :: Show Polls :: Message Navigator

Column based grammar [message #1075405]

Mon, 29 July 2013 11:18

Paulo Dias

Messages: 4
Registered: July 2013

Junior Member

I want to implement a grammar that uses the columns to validate the syntax.
For example, I want to validate that a specific keyword is written between the column 1 and the column 3 and that from the column 30 to the end of line it will be considered as a comment.
Is that possible using xtext and if yes can you provide an example?
Thanks very much,

Report message to a moderator

Re: Column based grammar [message #1075500 is a reply to message #1075405]

Mon, 29 July 2013 14:36

Henrik Lindberg

Messages: 2509
Registered: July 2009

Senior Member

On 2013-29-07 15:24, Paulo Dias wrote:
> I want to implement a grammar that uses the columns to validate the syntax.
> For example, I want to validate that a specific keyword is written
> between the column 1 and the column 3 and that from the column 30 to the
> end of line it will be considered as a comment.
> Is that possible using xtext and if yes can you provide an example?
> Thanks very much,

The easiest is to do this in the validator. Simply get the source
position from the node model and perform your validation based on that.

Regards
- henrik

Report message to a moderator

Re: Column based grammar [message #1075596 is a reply to message #1075500]

Mon, 29 July 2013 18:08

Paulo Dias

Messages: 4
Registered: July 2013

Junior Member

I think I do not explained correctly the issue.

I'm not talking about validation, but to have that rules enforced by the grammar. Explaining better (I hope so):

Is there a way to specify on a rule that expresses that a specific keyword is only upper case with 3 of size and starts at column 1 or that the comment can be any character starting at position (column) 30 til the end of line.

Thanks

Report message to a moderator

Re: Column based grammar [message #1075738 is a reply to message #1075596]

Tue, 30 July 2013 01:07

Henrik Lindberg

Messages: 2509
Registered: July 2009

Senior Member

On 2013-29-07 20:08, Paulo Dias wrote:
> I think I do not explained correctly the issue.
>
> I'm not talking about validation, but to have that rules enforced by the
> grammar. Explaining better (I hope so):
>
It is a really bad idea - because the user will be hammered with syntax
errors without further explanation or help. IMHO, you want to do this in
validation; that is almost the same as having the grammar enforce it. If
you want other features like code completion to behave differently
depending on where on the line it is invoked, you need to also handle
that naturally.

As a rule of thumb, the grammar should be as forgiving as possible since
this enables you to give the user a better editing/quick fix experience.

When you implement it this way, and something is in error, you can
provide the user with quick fixes like "move it/indent it, put it first
on the line, etc...".

> Is there a way to specify on a rule that expresses that a specific
> keyword is only upper case with 3 of size and starts at column 1 or that
> the comment can be any character starting at position (column) 30 til
> the end of line.
>
No.

The way to do it would be to enforce/check it in the lexer and emit
different tokens. This would only lead to grammatical errors which will
manifest themselves as syntax errors to the user.

> Thanks

Report message to a moderator