https://bz.apache.org/ooo/show_bug.cgi?id=126946

orcmid <orc...@apache.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |CONFIRMED
           Hardware|PC                          |All
         Issue Type|DEFECT                      |ENHANCEMENT
           Severity|Normal                      |Minor
          Developer|---                         |Research
         Difficulty|                            |
             Latest|---                         |4.1.2
    Confirmation in|                            |
     Ever confirmed|0                           |1
                 OS|Windows XP                  |All

--- Comment #4 from orcmid <orc...@apache.org> ---
(In reply to Lalith Ramesh from comment #3)
> > Manual link creation was also suggested as a workaround on the <a
> > href="https://forum.openoffice.org/en/forum/viewtopic.
> > php?f=10&t=82983#p384616">User Support forum</a>.  BTW, I'm using Gmail, and
> > it's smart enough to recognize both URLs.

> https://forum.openoffice.org/en/forum/viewtopic.php?f=10&t=82983#p384616

Thanks for that link.

I nosed around a little further and determined that automatic link creation is
an AutoCorrect feature, controlled at Tools > AutoCorrect Options > Options >
URL Recognition.  It is on by default.  

This applies across all components.  I confirm the described behavior in Writer
and Calc, not just Impress.

The reliance on the subdomain "www." appears to be by design and is consistent
with how web sites have been promoted in the past.

There is no "correct" approach here, so it is not a technical defect.  If the
AutoCorrect recognition of text that should be treated as links to URLs based
on the text is to be taken farther, that calls for enhancement of this
provision.

I have changed the Issue Type, The Hardware and OS, and the Importance.  

The Developer Difficulty is set to Research because we have the serious problem
of what the algorithm needs to be and how to prevent it producing too many
false positives.  It may be obvious when a human sees one of these that, in
context, it is known that a link is desired and what the actual URL needs to
be.  But creating more false positives that makes casual users (know how to)
turn them off is an issue to be considered.  Also, creating un-noticed false
positives during authoring is not a great thing.

How can we resolve this?

Also, note that even if we come up with a reasonable improvement to the
AutoCorrect case, no action will be taken until a developer determines this is
something that is worth their voluntary attention.

-- 
You are receiving this mail because:
You are the assignee for the issue.

Reply via email to