[Tcl-bugs] [ tktoolkit-Bugs-2191960 ] tests use identity encoding

[SourceForge.net <no...@so...>]

Bugs item #2191960, was opened at 2008-10-24 09:52
Message generated for change (Settings changed) made by dgp
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=112997&aid=2191960&group_id=12997

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: 52. [clipboard]
Group: obsolete: 8.6a4
>Status: Open
Resolution: Fixed
Priority: 5
Private: No
Submitted By: Don Porter (dgp)
Assigned to: Joe English (jenglish)
Summary: tests use identity encoding

Initial Comment:

A few tests in unixSelect.test
make use of the "identity"
encoding.

Test winClipboard-1.5 uses the
command [tcltest::bytestring]
which is based on the "identity"
encoding.

These tests are old.  Hoping there's
someone who can comment on what
they're meant to test.  Furthermore
whether the same testing objective
can be met using some part of Tcl
that we actually want to keep around. :)



----------------------------------------------------------------------

>Comment By: Don Porter (dgp)
Date: 2009-02-09 08:51

Message:
I'd have expected this to auto-close
by now.  Is that feature not active
in the Tk tracker?

----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2009-01-12 10:56

Message:

Should be tested by now.


----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2008-10-28 13:44

Message:

patch committed to HEAD so it
can be tested on Windows.

----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2008-10-28 13:21

Message:

Patch revises [testclipboard] command
on Windows so that it now directly tests
the internal routine TkSelGetSelection(),
which handles encoding details.  Besides
being a better test of code that's actually
part of Tk, this also means the test suite
no longer has to deal with invalid UTF-8
strings and can drop use of the ugly
[tcltest::bytestring] command.
File Added: 2191960.patch

----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2008-10-24 10:35

Message:

The unixSelect tests appears
to be checking that bytes
which are not valid UTF-8 can
be sent through the [selection]
mechanism without getting distorted.
These tests rely on the implementation
detail that [entry] widgets still
manage their contents as char arrays
and not Tcl_Obj's.  If that assumption
becomes false, there will need to be
a specialized testing command, it it
remains important to test this.

Something else appears noteworthy
about these tests.  I've believed that
[encoding convert* identity] was the
last remaining script level way to get
invalid UTF-8 into the objPtr->bytes
of some Tcl value.  These tests appears
to show that [selection get] can also
produce such a thing.  That's also a
bug according to modern standards.

----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2008-10-24 09:57

Message:

winClipboard-1.5 appears to use
[bytestring] just to produce
something to compare with the
result of [testclipboard].  A
revised [testclipboard] command
should be sufficient to squelch
that one.


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=112997&aid=2191960&group_id=12997