On 01/ 8/10 10:07 AM, Garrett D'Amore wrote: > Joerg Schilling wrote: >> "Garrett D'Amore" <gdamore at sun.com> wrote: >> >>> We have several device that are "live" on pcmcia, and have a lot of >>> complexity introduced to them via this. >>> >>> I strongly suspect there is even "dead" code, which we are keeping >>> around just because of Card Services API specifications. I hate >>> "dead" code. >> >> Eliminating so called "dead code" (if done decently) results in a >> rewrite >> and in a restructuring. This introduces new bugs and you have to put >> additional efforts into the project in order to fix these new problems. >> Looking at the long list of problems we currently have with Solaris >> in the multi-media area makes me believe that this is the wrong time >> for such a restructuting. >> This attempt looks like an effort to reduce the memory footprint of >> Solaris >> and there are other places where we could work in order to make Solaris >> run again on small devices. > > There have been several people "questioning" the merit of this > effort. While I appreciate your sentiments, I submit that its not up > to you to determine how I spend my time. In my spare time I work on > those things that are most interesting to me, or that I feel best > qualified to do.
All: The total answer is not up to the individual engineer. The decision to either do, or not do, this type of work has impacts beyond an individual's personal free time. Think; docs, QA, sustaining, C-Team, code reviews. There is very real cost, beyond just the 1 engineer, when doing any change. That's why it requires mgmt approval to decide if we invest in a specific area. That said, I do appreciate and agree with Garrett that PSARC does not evaluate investments and resourcing, so discussions here should be limited to architecture. Cheers, Neal > > Of course, you are also free to contribute where you see the the most > urgent need, or even just where you most want to. > > The questions I had raised about this case were "architectural > questions", and so far I've not heard anyone say anything negative > other than "why?" > > Note that the current cardbus framework is badly complicated by also > having to support pcmcia CardServices. I'd hope that we could > simplify the underlying nexus drivers and pccard framework with an eye > towards making the whole thing easier to sustain in the future. > > Anyway, I'll probably just drop this now... other things have caught > my interest. > > -- Garrett >
