Indentation Check with Namespace

[LupusUmbrae]

I've noticed that the indentation check seems to give incorrect results when indenting after namespaces. I'm expected with indentNamespace = true that the check expects code inside a namespace block to be indented

Take the follow code snippet:

  namespace pi {
    using rapidxml::xml_node;

With an indentation level of 4, and indentNamespace true i would expect that to pass, instead i get the error "Make this line start at column 1."

Later on when we indent again we get the same issue:

    PngDataSet::PngDataSet(...)
    {
        if (config.compare("") != 0)

With the above i get the error "Make this line start at column 5."

Thanks in advance

[jmecosta]

Yeah I think this check is somehow broken, we had some discussion about
this earlier already . ended up disabling it because cause to much noise

On Thu, Mar 26, 2015, 15:48 Robin Moss notifications@github.com wrote:

I've noticed that the indentation check seems to give incorrect results
when indenting after namespaces. I'm expected with indentNamespace = true
that the check expects code inside a namespace block to be indented

Take the follow code snippet:

namespace pi {
using rapidxml::xml_node;

With an indentation level of 4, and indentNamespace true i would expect
that to pass, instead i get the error "Make this line start at column 1."

Later on when we indent again we get the same issue:

PngDataSet::PngDataSet(...)
{
    if (config.compare("") != 0)

With the above i get the error "Make this line start at column 5."

Thanks in advance

—
Reply to this email directly or view it on GitHub
#441.

[LupusUmbrae]

Had a bit more of a look at this issue and it looks as though in my case the namespace declaration is the first node matched by translationUnit (AstScanner line 92), this means that it's the returned AstNode.

The top level AstNode is only passed in under visitFile, which in the indentationCheck doesn't check if it's a declaration that should cause the indentation to change level. Sadly i don't know enough to know if from visitFile i can determine what sub-types (i.e namespaceDeclaration) the translationUnit has.

Another way to do it would be to do a check on the first node, to check if it's inside a namespace body, i'll try this option and see if i can break it more :)

Hope that informations of some help :)

[guwirth]

@Typz code seems to be from you? Do you have time to fix it?

[LupusUmbrae]

Did some more investigation, my original thoughts weren't correct. It looks as though our namespace declaration is being parsed to a recoveredDeclaration tokenValue='namespace' tokenLine=27 tokenColumn=0 which isn't right.

After slowly cutting out bits an pieces of the code it looks like the issue is with >>, which i thought there was another issue in but can't find it at the moment

[markjohndoyle]

This one?
#49

[guwirth]

If there is an issue with >> try the latest 0.9.3. There is a bugfix #378
(please keep in mind that 0.9.2 needs SQ 4.5.2 or later).

[guwirth]

What should we do with this? Remove the check or fix it?

[jmecosta]

I ended up not having any use for it. Duno about others

On Fri, Apr 17, 2015, 20:28 Günter Wirth notifications@github.com wrote:

What should we do with this? Remove the check or fix it?

—
Reply to this email directly or view it on GitHub
#441 (comment).