Fwd: [GLLUG] Software Distribution

[Eduardo Cesconetto]

> I think I did not made myself clear, my suggestion is that whenever  
> there is a release, the release package should contain an installer  
> that would detect the system and pick the right binary to install,  
> just like Apple's UB.
> In my humble, the reason why RedHat and the Mac OS X are so widely  
> used and accepted are that for a developer to release to those  
> distros, there are a specific set of rules, that among other things,  
> guarantee(or try to) quality and stability. You rarely see that on   
> non-commercial distros, because developers are 90% basement geeks  
> that believe that their way to code should be the only way to code.  
> I have a good experience trying to be the "middle man" between the  
> customer and the developer, and most of the time, those are two  
> different animals that don't even speak the same language.
> Yes, FOSS is good and the OS model is great, but there should be  
> agreement to bring quality and uniformity, otherwise people will try  
> and get frustrated, and go back to(or stay with) Windows or Mac or RH.
> About not following a FOSS business model, if Clay can't close his  
> source, what stops Bill from using his commercial contacts to cash  
> into Clay's code? (names here are for illustration only)
>
>
>
> On Feb 17, 2008, at 6:24 PM, Karl Schuttler wrote:
>
>> Why not follow Apple's Universal Binary model?
>>
>> The universal binary was convenient as it carried users over from
>> PowerPC to x86, but beyond that, the universal binary isn't helpful  
>> on
>> any other architectures. ("It'd be nice to have some processor
>> independent executables on Linux.") 80% of developers are right; they
>> program for themselves or a company/organization that pays them to
>> develop something for specific architectures/flavors. Reducing the
>> number of linux "flavors" would not make the universal binary any  
>> more
>> relavent; one of linux's main points is that it can be ran on
>> everything from a toaster to a supercomputer (okay, so I haven't seen
>> one integrated into a toaster oven yet, but you just wait). Having
>> 'one' flavor is more beneficial to desktop users, but isn't necessary
>> for server side applications. You really need to fill up space  
>> running
>> a graphical interface, etc, on a server that will only be doing LAMP?
>>
>> Distros that make things easy for the user (Ubuntu, Suse, etc) are  
>> the
>> ones that have the most commerical potential. Why not do like other
>> developers and only build binaries for those that are popular and
>> represent your clientele? This seems to me what other developers have
>> been doing for years: developing for windows and mac OS but counting
>> out linux because linux users aren't the ones bringing in the dough.
>>
>> This response is less to Clay and more to Mike/Eduardo; I don't  
>> really
>> have any commercial software development experience, but I can
>> understand how building for multiple distributions could be
>> frustrating. All I ask is that you make the source available so  
>> others
>> can compile it; if you aren't, then you aren't following a FOSS
>> business model, which is why you are only releasing binaries. In this
>> case, developing for linux probably isn't your thing.
>> _______________________________________________
>> linux-user mailing list
>> linux-user at egr.msu.edu
>> http://mailman.egr.msu.edu/mailman/listinfo/linux-user
>