Menu
▾
▴
#42 mime::qp_encode failures (64 bit?)
On an Alpha XL-300 (64-bit platform) some of
the mime tests fail:
% source mime.test
==== mime-4.4 Test qp_encode with non-US_ASCCI
characters. FAILED
==== Contents of test case:
set str1 "Test de caractčres accentués : â î é ç et
quelques contrôles \"\[|\]()\""
mime::qp_encode $str1
---- Result was:
Test de caract=FFres accentu=FFs : =FF =FF =FF =FF et
quelques contr=FFle=
s =22=5B=7C=5D()=22
---- Result should have been:
Test de caract=E8res accentu=E9s : =E2 =EE =E9 =E7 et
quelques contr=F4le=
s =22=5B=7C=5D()=22
==== mime-4.4 FAILED
==== mime-5.1 Test word_encode with quoted-printable
method FAILED
==== Contents of test case:
mime::word_encode iso8859-1 quoted-printable "Test
de contrôle effectué"
---- Result was:
=?ISO-8859-1?Q?Test_de_contr=FFle_effectu=FF?=
---- Result should have been:
=?ISO-8859-1?Q?Test_de_contr=F4le_effectu=E9?=
==== mime-5.1 FAILED
==== mime-5.3 Test encode+decode with quoted-printable
method FAILED
==== Contents of test case:
set enc [mime::word_encode iso8859-1
quoted-printable "Test de contrôle effectué"]
mime::word_decode $enc
---- Result was:
iso8859-1 quoted-printable {Test de contr˙le effectu˙}
---- Result should have been:
iso8859-1 quoted-printable {Test de contrôle effectué}
==== mime-5.3 FAILED
Logged In: YES
user_id=75003
Maybe related to Tcl Bug 219223:
http://sf.net/tracker/index.php?func=detail&aid=219223&gr
oup_id=10894&atid=110894
Logged In: YES
user_id=79902
It seems mystifying to me; have you checked that it is not
an [encoding] problem? If the problem really is a 64-bit
nasty, it might be fixable with a quick [expr {...&0xff}] in
the right spots to trim all the offending bits...
Logged In: YES
user_id=80530
More data: I see these test failures in a Tcl 8.3.3
and in a Tcl 8.4a2 interp. I do not see them in a
Tcl 8.4a3 interp.
More magical corrections from the Unicode update?
Anyhow, a [package require Tcl 8.4] might eventually
take care of this.
Logged In: YES
user_id=75003
Priority down.
Logged In: YES
user_id=240
I don't see this bug anymore with tcl8.4. It does seem to
produce different results for the first failed test on x86
Test de caract=FFres accentu=FFs : =FF =FF =FF =FF et
quelques contr=FFle=s =22=5B=7C=5D()=22
and my linux ppc (unsigned char) machine:
Test de caract=C3res accentu=C3s : =C3 =C3 =C3 =C3 et
quelques contr=C3le=s =22=5B=7C=5D()=22
Logged In: YES
user_id=80530
If this package needs Tcl 8.4 to work
correctly. lets just add [package require Tcl 8.4]
and declare victory.
Logged In: YES
user_id=75003
Don confirms that this failure is gone for Tcl 8.4.2.
Logged In: YES
user_id=75003
Better solution, IMHO: Add known bugs section to mime
manpages and add info about this problem.
Forcing the large set of 32-bitters to switch to 8.4 for a bug
hitting only the small 64-bit crowd is not so good IMHO.
Logged In: YES
user_id=75003
Man page is extended now with section about known bugs,
and this item was filed.
Modding down, no other change for the 1.4 release.
Logged In: YES
user_id=80530
No test failures from mime 1.4 (tcllib 1.7).
This is fixed?
Logged In: YES
user_id=75003
Apparently. No test failures for me on a x86_64 platform as
well. Across Tcl 8.[2345].