Hi Shigio

Thanks for the quick respsonse.

On Thu, Aug 27, 2015 at 2:39 AM, Shigio YAMAGUCHI <[email protected]> wrote:

> I don't think such option is useful. Because something obtained
> with that is an incomplete one.
>

Incomplete, yes. If completeness of the entire directory tree was a
critical goal enforcing such behavior wouldn't be useful. But we are
talking about an option. Let's just assume that for the moment I might only
be interested in files of certain types which I know are all available but
I can't be bothered to create a manual list for.

Not my actual situation but a very similar example: Embedded OS kernel
alongside user mode applications and libraries. I know the common project
root and this is where I want to execute gtags. Whether I can access the
kernel and all other applications and libraries may not be interesting,
initially. Yet I want to tag one specific application plus all required
libraries, except I don't know which of the libraries are needed.
The project might contain legacy that isn't particularly pretty but an
unfortunate fact (welcome to the Real World), so the application I care
about may pull in code from another application, sidestepping a proper API,
etc. From compiling my application I know that everything I seem to need is
there, but I don't want to go through a compile log manually to pick out
the stuff I seem to need. I'd rather want gtags to support me (optionally!).

To get an appropriate permission bears a better result, I think.
>

Of course, but not always feasible. Lots of bureaucracy involved.


> By the way, gtags ignores orphaned symbolic links.
>

Good to know, thanks.

If you would like to skip unreadable files, you can use cp(1)
> like follows:
>

I know that this would be an alternative but not very convenient at all.
Sorry.

Best regards
Marcus
_______________________________________________
Bug-global mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-global

Reply via email to