From: Michael H. <mic...@ju...> - 2001-05-05 21:44:26
|
I've commited my first hack at the test suite in sdcc/support/regression. It is neither complete, tidy, well documented, or tested but that will change in the next few days. It also won't run on other peoples machines due to a couple of external dependancies. If anyone is interested in the format of the cases, see .../tests/*.c. Suggestions for test cases are also very welcome. Some of the features are: * Ability to run the tests on the host CC to validate the cases. * Hooked in through make's dependancies, so only failing tests get re-run. * Easy to add new ports to. A session currently looks like: sapphire:~/projects/sdcc/support/regression$ make Summary for 'host': 0 failues, 22 tests, 10 test cases Summary for 'z80': 0 failues, 6 tests, 10 test cases where 'host' is the suite running on gcc, and 'z80 is the suite running through sdcc and a Java based z80 emu I have. Ideally each port will report the same stats, but it turns out that the z80 mul/div functions are broken :) One problem at the moment is finding a cross-platform emulator for each port. I have no idea how to approach this at the moment. We do have the emu cores in the simulator, but I haven't investigated that yet. -- Michael |
From: Scott D. <sc...@da...> - 2001-05-05 22:39:09
|
On Sat, 5 May 2001, Michael Hope wrote: > I've commited my first hack at the test suite in sdcc/support/regression. > It is neither complete, tidy, well documented, or tested but that will > change in the next few days. It also won't run on other peoples machines > due to a couple of external dependancies. > > If anyone is interested in the format of the cases, see .../tests/*.c. > Suggestions for test cases are also very welcome. > Excellent! FYI, you probably know about sdcc/src/regression? There are just a hand full of files I created to test very basic functionality. I think your choice of placing the regression testing under support makes more sense. > One problem at the moment is finding a cross-platform emulator for each > port. I have no idea how to approach this at the moment. We do have the > emu cores in the simulator, but I haven't investigated that yet. gpsim is a PIC simulator definitely works under linux (and other unices). There has been a port to windows that may work as well. (the gui portion, which relies on the gtk port to windows, is very flakey, but the command line interface is all that's really needed.) Scott |
From: Sandeep D. <sa...@dd...> - 2001-05-06 23:51:52
|
Sorry missed Scott , maybe he would like to make the first release for the PIC port. How about it Scott ??? Sandeep -----Original Message----- From: sdc...@li... [mailto:sdc...@li...]On Behalf Of Scott Dattalo Sent: Saturday, May 05, 2001 3:36 PM To: sdc...@li... Subject: Re: [sdcc-devel]commit: Test suite skeleton On Sat, 5 May 2001, Michael Hope wrote: > I've commited my first hack at the test suite in sdcc/support/regression. > It is neither complete, tidy, well documented, or tested but that will > change in the next few days. It also won't run on other peoples machines > due to a couple of external dependancies. > > If anyone is interested in the format of the cases, see .../tests/*.c. > Suggestions for test cases are also very welcome. > Excellent! FYI, you probably know about sdcc/src/regression? There are just a hand full of files I created to test very basic functionality. I think your choice of placing the regression testing under support makes more sense. > One problem at the moment is finding a cross-platform emulator for each > port. I have no idea how to approach this at the moment. We do have the > emu cores in the simulator, but I haven't investigated that yet. gpsim is a PIC simulator definitely works under linux (and other unices). There has been a port to windows that may work as well. (the gui portion, which relies on the gtk port to windows, is very flakey, but the command line interface is all that's really needed.) Scott _______________________________________________ sdcc-devel mailing list sdc...@li... http://lists.sourceforge.net/lists/listinfo/sdcc-devel |
From: Scott D. <sc...@da...> - 2001-05-07 01:29:54
|
On Sun, 6 May 2001, Sandeep Dutta wrote: > Sorry missed Scott , maybe he would like to make the > first release for the PIC port. How about it Scott ??? The PIC Port is not ready. But I wouldn't let that impede the 2.2.2 release. Scott PS: The peephole -> pCode parsing is almost done. Look at this debugging stuff: movwf %2 str space pct % num movwf is mnem 1st operand is wild altpat_mnem1a movwf var 2 MOVWF %2 The first line is the raw text as found in the peephole file. The second line is "token" output of the parsed line. The second and third lines are the lexer recognizing the patterns. The 4'th line is a callback function that is invoked when a whole valid line has been recognized. The last line is the pCode->print() "virtual" function that is invoked after the pcode for the instruction has been created. I'm stoked! This has been a really tough nut to crack, but I'm starting to see light at the end of the tunnel. |