simple was added so that regression database didn't need to change along
with the changes to the television (simple is an older television
implementation rather than anything really new). but now that the
regress package can redux playback files, there is now no need for the
'simple' version
added Replace() to database package
simplified Regressor interface
playback regressor can now be reduxed
regression Redux() now uses database.Replace() instead of separate
Delete() and Add(). this makes sure that the redux entry gets the same
database key once the redux has completed
removed -dryrun option from REGRESS REDUX
removed regression fails log. it wasn't well developed and not a
particularly useful idea
fixed television.SetSimple(). the signal function was not set correctly
checks that linters exist before proceeding. project URL for each linter
in the Makefile comments
only go-errorlint and unconvert used at the moment
fixed all lint errors for the added linters
this is useful for recreating playback files for the regression
database. playback entries not currently supported for REDUX mode
added -playbackIgnoreDigest and -recordingFilename options for RUN mode
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
the information used to create the loader is storied in
the playback type rather than a created loader itself
this is more flexible and gives us better control of when
the loader is disposed of
checking of cartridge hash must now be performed explicitely
when the cartridge loader is created from the playback information
notifications interface instance moved to environment from
cartridgeloader. the cartridgeloader package predates the environment
package and had started to be used inappropriately
simplified how notifications.Notify() is called. in particular the
supercharger fastload starter no longer bundles a function hook. nor is
the cartridge instance sent with the notification
curated package predated the standard errors package introduced in
go1.13
the standard package does a better job of what curated attempted to do
the change of package also gave me a opportunity to clean up the error
messages a little bit
the reset can be done automatically but as an optional operation. it has
been discovered that in some situations more preparation is required
before the cartridge is reset. particularly, ACE/ELF type cartridges
where the ARM may executed before the coprocessor disasm/developer
packages are initialised and attached
the emulation package has been unecessary since the amalgamation of the
debugger and play modes. in order to allow switching between the two
modes it was necessary to remove the playmode package and to move all
playmode loops and other considerations into the debugger package. as a
result the abstraction offered by the emulation package is uncessary
this facilitates A/B testing of two binaries. normally you would want to
prevent a playback with a ROM binary different to one used in the
original recording, but in some instances it is useful
playback/recorder and driven input systems moved out of the the ports
package and into a new input package. how the input systems interact has
been clarified and improved - for example, it is now posssible for a
playback file to be used to drive two emulations for comparison purposes
the debugger startup procedure has been clarified with two distinct
startup functions for playmode and debugger - each of which take
different arguments. the clarity has allowed the reintroduction of
recording and playback to the main play mode
update userinput.EventKeyboard to support Repeat keypresses
in playmode rewind forward/backward occurs on shift-left or shit-right
cursor keys (with repeat)
removed the pause feature (sdlimgui) that caused two-frame kernels to
appear "complete" when paused.
a) it wasn't consistent
b) it can cause single-frame kernels to shimmer
c) it doesn't really work with the rewind system
it'll be addressed in a future commit
makes it more convenient/clearer when working with all three coordinate
values (frame, scanline and clock)
changed all packages to GetCoords() where appropriate
regressionDB wasn't storing the requested cartridge mapping correctly.
for example, if no mapping was specified by the user but a mapping was
detected by file extension then the detected mapper was stored. this
caused problems with supercharger audio files - the AR mapping was
stored in the DB but no that it was an audio file
DB now stores the user specified mapping in all instances (ie. not the
mapping detected by the file extension).
related to this was that the deserialisation of DB entries was not using
the cartridgeloader package correctly. instead of using NewLoader, the
deserialisation was populating the Loader type manually.
this has revealed the way forward for an effective convergence of
playmode and debugging mode. for this commit a Playmode interface has
been added but this will abstracted into an Emulation interface in the
future
this is marginally quicker but I feel it's neater. to solve the problem
of distunguishing between returns values (true, nil) and (false, nil)
the latter can implemented with a sentinal error.
it's less clumsy when used as a label and it's accurate with regards to
how the VCS is documented. not sure why I thought HorizPos was a better
label. maybe it helped my thinking in the earlier phases of development.
this affected regression tests when random state was switched on
- although regression tests reset the preferences to a known state
the timer was not affected when the VCS was reset, causing some
tests to fail inexplicably
