specifying FAILS as a key will add the previously failed regressions to
the list
REGRESS LIST accepts list of keys to list. this can include FAILS to
list the previous FAILS from REGRESS RUN
as before, specifying no keys at all defaults to selecting all keys in
the database
since moving the ARM clock setting to be preference driven
latency and cycle stretching factor for flash access was performed
too early. this resulting in a stretching factor of zero
bug introduced in 8e4838d. this affected v0.31.0
there's a lingering question regarding the PCLK division. the default
value for the architecture is a division by 4 but I think the Harmony
driver probably sets it to a division of 1 before handing over to
the custom code
it is set to 1 for now because historically that is the value we've used
and measurements indicate that it is a correct assumption
this could happen when changing directories when a previous list was
ongoing. this was supposed to have been fixed in fcb8439 and 02bb6a33
but both were incorrect in some sitations
the new solution correctly waits for a cancel to complete using a mutex
this runs 'go vet' but filtered through awk so we ignore noisy errors
that we're not interested in
fixed vet error caused by unreturned error type in input package
changed comparision/diff texture to linear and clamped scaling
comparison window will not be open on following session
if a comparison window was left open on program exit, it remained
open when the program is started again, even if there was no
comparison ROM specified. comparison window will now be closed
under those circumstances
the test now affects whether the wait channel is used. the frameQueue is
not affected directly
this fixes problems with the screen not appearing to update (in
playmode) when the TV is of a higher refresh rate than the monitor. this
happened intermittently depending on where the frame queue indexes where
at the time the TV settled on the problem frequency, so it was difficult
to debug. however it happened more-or-less consistently when running the
Todd Frye version of xevious
setting the spec in this manner changes the creation Spec ID. this means
that is the spec is changed by the TV SPEC command or by via the debugger
window, newly loaded ROMs will prefer the creation ID. if the new creation
ID is AUTO then auto-detection and filename directives are preferred
changing the creation ID to AUTO with the forced flag also resets the
stable frames counter. this means that the auto-spec identification will
work as expected on change
for example, the ROM selector created a new thumbnail animation with
the AUTO spec when it's more appropriate to use the creation spec of
the main emulation
tracker package uses television spec ID correctly
movie cart tested for PAL60. it should be PAL-M
rewind search functions uses the TV spec of the main emulation
command line "performance" and "regress add" -spec option renamed to -tv
template for TV terminal command uses specification.ReqSpecList
this is instead of swapping FrameInfo information and setting the
resizer values based on that. this is used by the television in
conjuction with the preview package
this fixes a problem where some ROMs when given the command line
directive to use the PAL spec is contradicted by the preview emulation.
this seems to be a particular problem with PAL ROMs that never set a
valid VSYNC signal. Nightstalker is a good example of such a ROM
there was a very small % chance that the List() would end before all
entries had been received
'go test -shuffle on -count 100000 ./archivefs' reports no failures
