* Zero3 <zero3 at zerosplayground.dk> [2008-11-26 22:07:11]: >>>>>>> Next, we must identify anything that can be improved in Freenet that >>>>>>> would make writing these installers easier. >>>>>>> >>>>>> IMHO moving the "wizard" part into the node itself was an >>>>>> important step in the right direction. We could move the rest >>>>>> into the node by always downloading the plugins and seednodes >>>>>> file in the installer, and asking the user about the plugins >>>>>> during the post-install wizard. Ideally we'd also ask the user >>>>>> about auto-start in the post-install wizard (defaulting on but >>>>>> executing a script to turn it off if the user asks us to). >>>>>> >>>>> I agree. It doesn't seem like that big of a task to move the rest >>>>> of the stuff into the wizard (now you already have the >>>>> framework). >>>>> >>>>> >>>> Putting stuffs in the wizard goes against the packaging logic. On debian >>>> you would want to use debconf to ask the user on how to configure his >>>> node... >>>> >>>> >>> Both ways should probably be supported. >>> >> >> I see... and how exactly is that going to reduce the maintainance cost >> again? I must have missed something. >> > > It's more about "doing things right", really. Obviously that does > require extra work in the beginning, but if you look at the big picture, > I think you will save time in the end by unifying the installation > procedure with the other software in the world. I won't try to convince > you about that if you disagree, because that would kind of be even more > a waste of time. >
You don't get the picture: we are facing a human^wcodemonkey resource shortage here. There is no need to convince me that each packaging system has its assets and that having a properly written, maintained package for each of them would be great... I am convinced of that... But I do know that they are other areas of the code that needs improving too. Not to mention that packaging (and more globally speaking dealing with installation/platform specific issues) is everything *but* fun. That's why I am not spending any more of my time on those related matters. Right now we do provide and maintain *one* cross-platform installer; It's arguably not perfect (hehe...) and does have defaults. It could use some improvements... but no one is willing to work on that. I do not think that it's realistic for the project to spend any time attempting to provide packages for any platform. Really that job needs to be outsourced. I am one of those who think we shouldn't even provide binary builds but source code. Many projects do it this way... and they do find people to build and package their code. I will write some documentation on how to build and distribute freenet in a cross-platform way. Then it will be up to people to follow the guidelines and package the software. My guess is that writing that documentation will be a waste of my time... but well... Let's try to be optimistic for once. So far no one has even replied to Ian's call for help (http://archives.freenetproject.org/message/20081124.201152.ab6aeaba.en.html). -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20081126/225cb956/attachment.pgp>
