On 5/12/2017 11:05 AM, Ross Berteig wrote:
This patch is identical to an experiment I ran last night. It gets
past the obvious error, but still fails with a configure command that
worked before:
$ ../fossil/configure --with-zlib=tree --with-openssl=/local/ssl
--json --with-tcl --with-tcl-stubs --with-tcl-private-stubs
--with-th1-docs --with-th1-hooks
Host System...i686-pc-mingw32
Build System...i686-pc-mingw32
C compiler... cc -g -O2
C++ compiler... c++ -g -O2
Build C compiler...cc
Checking for stdlib.h...ok
Error: Option --with-tcl requires a value
Try: 'configure --help' for options
After tweaking my configure command options so that --with-thing became
--with-thing=1, configure runs without errors. I think that command is
still compatible with the older autosetup, so I don't mind changing it
in my tooling even if autosetup gets another patch. I just want to be
able to bisect across this change in autosetup without needing to make
my build tooling aware of the fossil version.
Putting my test maven hat back on, I did let all the tests run from tip
of trunk just now.
Fossil configured as above passed all tests.
So did fossil with a minimal ../fossil/configure --with-zlib=tree
--with-openssl=none
I did find one minor test framework glitch. Apparently last march I was
flailing at some test case and was running the test framework with
-keep, but letting the folders accumulate in the usual %TEMP% place. On
Windows this is C:\Users\Ross\AppData\Local\Temp and it doesn't get
emptied automatically by much of anything. So there were some folders
named home_* and repo_* left behind, with random content. The balance of
the name is a process ID which is of course reused from time to time.
Apparently reused today.
This caused the test framework fits as it was cleaning up tests, as it
only seems to delete what it knew it put there in the first place. So
every single test file caught an exception from the framework attempting
to remove a non-empty folder. I think one could argue that the test
cleanup should remove the entire home_ and repo_ folders that it created
(or found already there) without concern for anything extra that
happened to be there. I haven't chased that down in tester.tcl, but I
admit to being tempted to find the right spot and make that change.
--
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