|
RA is not itself an emulator, but rather an open-source framework composed primarily of multiple emulator "cores" commonly ported from independently-written standalone implementations and sometimes enhanced by the RA development team. RA serves as a reference implementation of "libretro", itself a generic API that allows RA not only to be core-agnostic but to extend beyond emulation purposes as a flexible and powerful instrument of its own right.
RA's main benefits include its availability on many host platforms; its support of many "best of breed" emulators under a single framework; its streamlined interface which eschews keyboard/mouse navigation in favor of controllers and other simpler means of input; and its support of cross-platform "shaders" which lend visual authenticity when used properly.
That said, RA's primary contribution isn't the preservation of old technology for historical purposes (this site's focus); but rather on broadened awareness and usefulness of such technology to a wide audience across an expanded range of platforms such as personal computers, tablets, and whatever the future may hold. RA could achieve certain preservation goals indirectly however, given the theory that emulation of individual target systems stands a greater chance of being preserved as part of a collective framework than on their own, as history has shown. This is of course similar to MAME in concept, though RA's approach and design differs in several fundamental and obvious ways.
While RA may offer its users certain key benefits, we encourage the use of its upstream emulator counterparts wherever possible, given several of RA's negative aspects and limitations:
-
Several developers of emulators that comprise much of the code within downstream RA cores have become frustrated with the treatment they've received from the RA team, and have aired complaints such as lack of proper attribution and credit, issues with licensing, or other similar matters. Sadly, as a result some of them have chosen to stop collaborating with the RA team, close down their own projects, or in the worst cases, close their source. This runs counter to NonMAME's goal of encouraging the kind of openness and information sharing that leads to more accurate emulation and better system preservation in turn.
-
RA distributes and promotes old MAME builds such as "MAME2003" which harm the goals of both the MAME team and preservation in general. While the RA team's stated intent is performance, given that such old builds run faster on older hardware, such builds lack the tremendous progress MAME developers have made over decades, not only to improve the accuracy and experience of emulating these old systems, but to add literally thousands of new target systems. It is not difficult to see how dispiriting and demotivational this has been to the MAME team, as their hard work and vast improvements are unrecognized. Furthermore, old MAME builds require old ROM sets, many of which contain "bad dumps", which are then widely circulated within RA's user community. This directly conflicts with system preservation goals that demand precision and accuracy in the target system. With the above in mind, we feel RA should support only the most current MAME build.
-
RA's cores require constant maintenance to keep up to date with upstream code changes. It's common for development work on a given core to slow or stop altogether, and in such cases the upstream project quickly renders it obsolete. Upstream developers could certainly lend a helping hand, though there seems to be lack of motivation due to the issues stated above. We believe that with a more friendly, respectful and collaborative working environment between RA and such projects, these kinds of issues could be more efficiently resolved to the benefit of both parties.
-
RA seems to suffer from "scope creep" that offers little benefit to users while distracting developers from more targeted and valued improvements. For example, the team expends actual development effort getting RA to run on such archaic systems as Windows 95 and the PlayStation 2, where focus on current desktop and mobile device-based operating systems would promote more efficient development. RA also supports an enormous volume of emulator cores that only grow over time, when they could instead focus on a pared-down set of key cores with greater effectiveness. For example, dropping support for all NES cores other than the most accurate would permit greater focus on that core. Other features of questionable benefit have been added to RA over the years as well. The resulting top-heavy maintenance burden could be significantly relieved to benefit users and developers alike.
-
Many cores lack all the features of the original upstream emulator. For example, RA cores tend only to emulate computers in a way that lets users play the games available on them, rather than allow full exploration of the system and its associated peripherals as the original upstream emulator does. This reflects a design philosophy that is at odds with system preservation goals.
-
If the input of an emulated system doesn't fully conform to RA's generic retropad, keyboard, or mouse mappings, some input parameters are removed in the RA core. Ideally, if an RA core's input interface cannot be retroactively mapped using one or more of the three generic RA mappings, users should be allowed to map the missing inputs themselves.
-
RA's MAME core only fully supports MAME software lists in limited cases such as the NES, so its usefulness is somewhat limited. Along the same lines, support for proper software dumps, such as the "split-ROM" format used by MAME's Nintendo 64 and Super Nintendo drivers, would also boost its usefulness and further its accuracy goals. In fairness, RA recently added support for the MAME CHD format in many key cores.
|
