>> Karel> Why do you need to maintain your hwclock separately? I don't see a real >> Karel> technical reason. >> > Bela> The historic reason is that an old version of hwclock was captured into > Bela> util-linux long ago, and not maintained in parallel with the separate > Bela> hwclock project that _already existed_ at that time. > Karel> Are you really sure? Karel> Karel> The original clock(8) (Apr 1992) has been always distributed within Karel> util-linux. Brayn's hwclock(8) (Sep 1996) -- based on clock(8) -- has Karel> been already available in util-linux-2.7 (1997). The history is probably not relevant to whether there should be two hwclocks today, but let me clarify it a little: 'clock' by Charles Hedrick is almost as old as the Linux kernel. It was one of the original components of util-linux. util-linux in those days was much more of a distribution project than a development project, and its components (well at least some of them) were maintained by upstream maintainers. Charles and other developers of 'clock' stopped maintaining it (I don't think they ever officially quit; they just stopped), and as a result, it didn't move forward much for several years. I was one of the many people who were frustrated with a need for changes to 'clock', so in 1996, I made my replacement, 'hwclock'. The code was heavily derived from 'clock' and backward compatible, but also massively rewritten because the primary new feature is that it did subsecond precision by waiting for and observing the moment of the clock tick. That required different structure. I knew other people would appreciate the enhancements, so I made 'hwclock' available to the public, but I knew that most people frustrated with 'clock' would never know it existed and keep suffering. There was no Sourceforge, Freshmeat, Google, personal web servers, etc. So I suggested to Nicolai, the maintainer of util-linux, that he include it in the package, where people were likely to notice it. Nicolai was already bothered by the problem reports about 'clock' that he couldn't do anything about (because remember, his role was mainly to distribute), so he went a step further and declared 'clock' obsoleted by 'hwclock' and stopped distributing 'clock'. So util-linux obviously became the main source of 'hwclock'. However, I was still the maintainer. Users usually sent their requests and suggestions directly to me because the documentation said I was the maintainer, but sometimes they sent them to the util-linux mailing list and Nicolai forwarded them to me. I regularly sent new releases to Nicolai, who packaged them into the next release. That changed in 1999 when a new util-linux maintainer, Andries, decided to stop taking new versions from me. It would take a few paragraphs to explain why, and this is long enough already, so I'll just say that after that, I continued developing hwclock and many people started getting it directly from me. So: when Bela speaks of hwclock being "captured" into util-linux, that could accurately describe the fork that happened in 1999. But there was no hostility involved in hwclock joining util-linux in the beginning. It was my idea. The original question was what the technical reason is for distributing a hypothetical single hwclock independently. I don't know what a "technical reason" is here, but I just happen to believe that smaller engineering projects are more efficient and flexible than bigger ones. For the same reason I divide a program into modules, I like to divide a development project into self-sufficient pieces. -- Bryan Henderson San Jose, California -- To unsubscribe from this list: send the line "unsubscribe util-linux-ng" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
