Since a lot of the dust has settled in the fossil 2.0/2.1 and SHA3
world, I decided to build version 2.0 configured with JSON, TCL, and SSL
run the test suite. I build on Windows 10 Pro 64 bit, using GCC under
MinGW's port of bash, aka MSYS.
I found a regression in wiki.test: wiki-41 fails because fossil wiki
export filename.txt --technote abcdefabcdef fails if the hash is not a
complete hash. We always used to allow a hash prefix there. This
specific test case is verifying that a hash prefix is allowed. This
appears to result from the introduction of hname_validate() in checkin
[5d8abf92].
There is also a bug in the test case, in that it didn't verify that the
fossil command succeeded, and that caused a TCL error to be thrown when
attempting to compare file content to a missing file, halting the test
suite badly. With that bug fixed, we find the regression caught by
wiki-45 and wiki-46 as well.
There is another regression triggered by the JSON test suite. An
assertion fails when using fossil sqlite3 to create a deliberately
broken repository with which to trigger a specific JSON error condition.
That code has worked fine since it was last touched a year ago:
write_file bad.sql {
CREATE TABLE spam(a integer, b text);
}
exec $::fossilexe sqlite3 --no-repository bad.fossil <bad.sql
#fossil_json HAI -R bad.fossil -expectError
It does appear that the only reference to bad.fossil is the commented
out line following its creation. We should either actually find a way to
stimulate FSL_JSON_E_UNKNOWN and then test for it, or not create
bad.fossil at all. With the dead code in fossil.test removed, that test
case now runs to completion.
I don't see an existing test file that exercises the fossil sqlite3
command, so I'm not sure right now where I should stuff a better test
case for whatever regression is being triggered. I do see other tests
that use fossil sqlit3 to build repositories from scratch with specific
artifacts.
I've got a quick hack fix for both test cases which I just checked in in
the rbk-2.0-tests branch.
--
Ross Berteig r...@cheshireeng.com
Cheshire Engineering Corp. http://www.CheshireEng.com/
+1 626 303 1602
_______________________________________________
fossil-dev mailing list
fossil-dev@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/fossil-dev
