Hi On Tue, Jul 22, 2014 at 6:15 PM, Jeremy Bennett <[email protected]> wrote: > The last time there were some reasonable gripes (lack of git support, > lack of mailing lists). But it divided the community and led to a lot of > upset. A few years before that a group split off to form BA Semi, and we > now see the problems there as we try to upstream GCC, and find that some > of that group were contributors.
I'm not trying to alienate anybody. Honestly, it might be too early to say this but I don't see this being a problem. > But this is a diverse community. Some of us do use OpenRISC for living > (we were paid for the GCC 4.5.1 work). But that revenue allows Embecosm > to sponsor ORConf - like lots of open source companies we do give back. Of course, I'm not debating that. I love that people work with OpenRISC and are part of the community, I'm just stating that I don't have chain of command or financial interest to do this - I do it because I believe it's the right thing and will help the project move forward. Please don't think I don't approve of people earning money by using OpenRISC. > OpenCores problem has always been a lack of revenue. It is why Damjan > sold the website to ORSoC AB several years ago. And ultimately the > support they can give depends on the revenue they get. That might have been true in the past, but I know there are better ways to tackle this today. Crowd funding and donations comes to mind. > Those are fair complaints, but the approach would be to talk to ORSoC > AB. I remember the horror of the 18 months before ORSoC AB stepped in. > The website was always down, and the website was completely unusable. > You may not like ads, but they do pay for the servers and some maintenance. I would love to see some receipts or books that tell me that this is the case. >From what I've seen, the website is in terrible shape - see below regarding >SVN. > Are you going to keep them in sync? On a best effort basis. The idea here is more to make a stand, and that voices in the community has asked for it. It was also evident that this is needed, a lot of data from OpenCores SVN is not accessible, either due to configuration errors (timeouts) or plain data corruption (CRC errors). With this effort at least we have a backup. > Did you ask the project managers of > those 800+ cores if they thought this was a good idea. I think you know the answer to that question. Obviously I did not, and as free software tends to be "the one with community support survives" I'm not troubled by this. Time will tell if this was a good move. > I know many of > them are dead projects (as are many on GitHub, and on SourceForge), but > there are also some that are not, even if they are not projects you use. I count on that. My some-what educated guess is that the clear majority are dead or very seldom updated. In this case it allows the community to actually patch the projects on FreeCores through pull requests on github - instead of staching .patch-files somewhere like orpsoc-cores does today for example. > What about the discussion forums (including the non-OpenRISC ones). Are > you migrating those as well? That is often where the beginners first > come in. I have no objection on leaving this on OpenCores. I'm not asking about a blind abandonment, just moving whatever is critical and making the initial user experience better. > I don't want to submit a pull request that relies on someone else to > decide. I want a wiki which I can edit. If you don't like the current > wiki, just edit it. I was a bit non-clear here. The goal with the website repository is not to replace the Wiki, but replace the current http://openrisc.net and gather information that is stable enough on a more friendly, structured and reviewed site. The wiki serves a purpose (scratchpad for ORconf, writedown of a tutorial) which I only think another Wiki would solve, so for now I don't propose replacing it. In essence, I want to move http://opencores.org/or1k/Main_Page, http://opencores.org/or1k/OR1K:Community_Portal and a few other pages to a more stable format, while having links to the Wiki for other more "exotic" stuff like workshop results / orconf things. > We have now just given up. I generally propose just doing it if talking fails. I don't think people care, they just want a simple interface. > I can easily see this becoming a *third* mailing list. The only thread that was not sent to both lists during 2014 was 'deprecated compiler name?'. Probably because OpenRISC is one of the few projects that has two mailing lists, and that is confusing. Shutting down one mailing list and making it bounce with a message telling to use the other will work just fine. > Don't underestimate the work required for a community of this size. Just > looking at the OpenCores OpenRISC mailing list since last October, there > have been around 450 posts, with 42 different contributors, most of whom > are multiple contributors. At the same time there have been 220 forum > posts, from another set of contributors. I don't see how this is a problem. I've requested an extraction of the mailing list member list, if I get an answer for that - I'll simply add them to the other mailing list. That happens all the time, at least in my experience. The big problem here is that OpenCores admins are unresponsive, which I address last in this email - and ironically is one of the reasons why I want to move critical stuff off the platform. > It seems that we currently have OpenRISC hosted on GitHub, which gives > the version control the community desires. Do we really need to split > everything else off and do it for every project. GitHub for hosting > repos is good. This is a big point, but as I see it we've already done this. All the major engineering doesn't go through OpenCores. > Its website/wiki stuff is so-so. Agreed, it is however free, community trusted and very robust. A stable website like what I'm proposing would survive hosting moves, if the current Wiki dies we only have the occasional backup. > Its issue tracking is a nightmare for any project with more than 100 lines of > code. Agreed. But I cannot say that Bugzilla is any better. In this case the issues are at least categorized into the component that is relevant (which admittedly can be a problem). We're already using github issues for some stuff, so it's more going with the path of least resistance here. > And is this really what you want to do? Surely your interest is in > engineering. Clearly you underestimate my passion for tidyness :-). This endeavor is fueled by the same energy that I use for upstreaming. > I guess having been through this twice before, I really don't want to > face the pain. Could I suggest you at least hold fire until ORConf when > you can talk to all the main OpenRISC contributors. With all due respect, no. I don't think having IRL meetings will solve this. IRL meetings have the added overhead that everybody that took his/her time to come there has the psychological pressure to add his/her opinion to shape the decision and might not have the full picture. The awesomeness of open source and the Internet is that it should be preferred to make decisions online, and respond quickly to winds turning. > That gives time to discuss what the solution to the obsessive desire of > OpenRISC to implode on a regular basis is. I am not convinced that > another individual splitting off a fork is going to be any happier > experience than the last two efforts. To be clear: I'm not proposing moving much. I'm proposing de-coupling, removing dependency on OpenCores. As I already said I don't see any point in removing the Wiki for now, the forums will stay, and the development workflows we have today are unchanged. With this I'm trying to set a standard for the project to not use whatever is best for whatever we're trying to do. Sadly, OpenCores fails for that when it comes to source code and information portals. > It may be time to go down the route I have long advocated, which is a > proper community organization, not controlled by any one individual. > Given ORSoC AB have changed their CEO in the last year or so, you might > even find they would back this now. The advantage of a properly > constituted community organization is that you are much more likely to > persuade corporate sponsors to support you. The OpenRISC organisation is "owned" by a couple of active maintainers. I think that's the best solution. Having a real corporate entity "own" stuff just makes it harder. It might be useful if it comes to actually seek funding, but I would vote for always have the control in the community, never a formal organisation in a single country. That is however I think a separate topic. Now for ORSoC. Is this still the name of the company? orsoc.se just redirects to kncminer.com. All older orsoc.se links I find are gone, including documentation that was once available. Any attempt I've made to contact the maintainers of OpenCores (which I assume is ORSoC) has remain unanswered. Someone is probably still moderating the forum, but that's as far as it goes for life signs. I feel very uneasy about this. For what it's worth, I've been contacted by numerous people off-list that think this is a great idea but asked to be anonymous. I appreciate your input as always Jeremy, and you have much more experience with the community in these matters - but I think the current approach is trying too hard to please everyone and ends up going nowhere. Regards, Christian _______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
