On 1/11/2016 1:15 PM, Joe Mistachkin wrote:
Ross Berteig wrote:
* th1-tcl-2
* th1-trace-2
* th1-trace-4
* th1-info-commands-1
* th1-info-vars-1
* th1-info-vars-2
* th1-info-vars-3
* th1-info-vars-5
The "th1.test" file should now fully pass on trunk, even when the Tcl
integration and TH1 hooks features are enabled via the environment.
And they appear to all pass, with the th1-tcl-2 quirk understood below.
For the "th1-tcl-2" test, I assume it failed because the SQLite package for
Tcl is not available? Running "th1-tcl.test" with the -verbose option should
tell us the exact result.
Bingo. I am not a frequent user of Tcl these days, and have only
whatever amount of TCL comes in a stock Ubuntu install on this VM. (I
added tcl-dev and tcl-docs to that while flailing at getting fossil to
compile --with-tcl, but that probably wasn't needed for compilation
--with-tcl-private-stubs.)
Running th1-tcl.test with -verbose shows me that it skipped th1-tcl-2,
saying "SQLite package for Tcl not available".
So I did "sudo apt-get install libsqlite3-tcl" and now it runs and
passes th1-tcl-2.
can't set "data(0)": variable isn't array
while executing ....
Also fixed on trunk.
It also "fixed" the merge-utf-27-23 and merge-utf-27-32 tests. Not
because a bug was found and fixed in the merge code or those tests
specifically, but because they were using the content of test/utf.test
and changing that content changed the outcome.
I'm slightly nervous about that, it has a "fixme" sort of smell about it
and I wish someone already versed in the merge code and its test cases
could evaluate that failure and reassure me. You'll need to make a
free-standing case that explicitly uses the version of utf.test (and
possibly the matching version of merge2.test) from before checkin
[ad6de63833] in order to diagnose it.
* th1-checkout-2
* th1-globalState-1
* th1-globalState-8
I'm not sure about these. These are using the "run_in_checkout" procedure,
which should allow them to pass as long as "tester.tcl" itself is being run
> from within the Fossil source checkout.
They pass now, but this is a build with all the TCL and TH1 features
enabled. Maybe those tests aren't selective enough and are really
testing if a feature is present?
For the full test suite, and fossil configured --with-openssl=none
--with-tcl --with-tcl-stubs --with-tcl-private-stubs --with-th1-hooks
--with-th1-docs, I have a only 2 errors out of 29336 tests. The errors are:
* merge_multi-4
* merge_renames-5
For the full test suite, and fossil configured --with-openssl=none, I
have 5 errors out of 29300 tests. The errors are:
* merge_multi-4
* merge_renames-5
* th1-checkout-2
* th1-globalState-1
* th1-globalState-8
I think that is suggestive that the three th1 tests that failed are
failing due to something that changes when the Tcl and TH1 features are
configured in and out, something that relates to what "checkout",
"{globalState checkout}" and "{globalState repository}" evaluate without
Tcl compiled in.
--
Ross Berteig r...@cheshireeng.com
Cheshire Engineering Corp. http://www.CheshireEng.com/
_______________________________________________
fossil-dev mailing list
fossil-dev@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/fossil-dev
