Tcl Source Code

The attached patch, plus the three new files, doc/lset.n, tests/lset.test and tests/lsetComp.test, 
comprise the reference implementation for the functionality described in TIPs #22, #33, and #45.  
It is posted here for review by the affected maintainers.

Logged In: YES 
user_id=72656

This was tested and commited to 8.4a4cvs on Nov14.

Logged In: YES 
user_id=80530

Latest patch applies clean to HEAD and 'make test'
reveals no new failures.  I'm trusting the developers
to have come up with a complete test set.

One change I would make to the docs in lindex.n and
lset.n.  They have syntax like:

\fBlindex \fIlist ?index...?\fR

but then the description that follows refers to
the "\fIindices\fR" .  I think that would be better
phrased the "\fIindex\fR arguments"

Note I think there are also several instances of 
\fP in the man pages where \fR is intended.

File Deleted - 12135: 



File Added - 12923: tip-22-33-45.patch4

Logged In: YES 
user_id=99768

Fourth version of the patches.  Fixes a bug that DKF discovered
where it was possible to assign to ptr2 in an object's internal
representation without first ascertaining that the object was
a list.

Logged In: YES 
user_id=99768

Fourth version of the patches.  Fixes a bug that DKF discovered
where it was possible to assign to ptr2 in an object's internal
representation without first ascertaining that the object was
a list.

File Added - 12322: lset.n

Logged In: YES 
user_id=99768

At dkf's suggestion, added to lset.n a paragraph describing the behaviour when a list index is out of range.

File Deleted - 12059:

Logged In: YES 
user_id=80530

'make test' with the latest revision:

==== obj-31.5 regenerate string rep of "end--bigInteger"
FAILED
==== Contents of test case:

    testobj freeallvars
    teststringobj set 1 end--0x80000000
    testobj convert 1 end-offset
    testobj invalidateStringRep 1

---- Result was:
integer value too large to represent
---- Result should have been (exact matching):
end--2147483648
---- Test generated error; Return code was: 1
---- Return code should have been one of: 0 2
==== obj-31.5 FAILED

File Added - 12135: tip-22-33-45.patch3

File Deleted - 12098: 

Logged In: YES 
user_id=99768

On the chat, dgp objected to the fact that tclEndOffsetType wasn't a registered type and lacked an 
"update string" procedure.  The third version of the patch adds the missing features.  It also adds an 
"invalidateStringRep" option to the [testobj] command, so that the "update string" procedure can be tested: 
otherwise, the string rep of an "end-offset" object is never spoilt intentionally, nor is such an object ever 
built except from a string.

File Added - 12099: lsetComp.test

Logged In: YES 
user_id=99768

Also removed a stray 'exit' from lsetComp.test -- how'd THAT get in there??? 8^)

File Deleted - 12061:

File Added - 12098: tip-22-33-45.patch2

Logged In: YES 
user_id=99768

tip-22-33-45.patch2 is a revised set of patches for the modified files.  It contains two changes from the 
original:

(1) a memory leak in tclCmdIL.c (TclLsetList) has been plugged

(2) the symbols TRUE and FALSE (defined on Windows, not elsewhere) used in tclListObj.c have been 
replaced with 1 and 0 respectively.

File Deleted - 12058:

Logged In: YES 
user_id=80530

generic/tclListObj.c does not compile
because TRUE and FALSE are not defined.

File Added - 12061: lsetComp.test

File Added - 12060: lset.test

File Added - 12059: lset.n

File Added - 12058: tip-22-33-45.patch