> [Original Message] > From: Geir Magnusson Jr. <[EMAIL PROTECTED]> > To: <harmony-dev@incubator.apache.org> > Date: 10/1/05 11:10:55 PM > Subject: [project policy] Author credit and attribution > ...snip... > > I've worked in projects that did it, and some that didn't. When tags > were there, I think it gave people a chance to 'sign' their work, and > I'll be the first to admit that when I did my first-ever commit that > had my name on it, I was proud! It's a natural thing to be proud of > our work. The flip side was that I've seen it lead to people > believing they "own" a piece of code because of the tags, I've seen > "keeping up with the joneses" where every contributor adds an author > tag, no matter what, leading to strange feelings about what is the > level that makes on an "author".... For example, reformatting w/ > eclipse? >
SOAPBOX #1 ---------- I have always found author tags useful. For a distributed work environment like the ASF, not everyone has immediate access on a LAN to the source code repository, and I have always found it useful to be able to look immediately at a given source file and know _immediately_ who made the last change. If I have been paying attention to recent activities, a simple author tag tells me not only who made the last change, but very likely what that change entails. On RCS- and CVS-based systems, the log tag also provides the change history, so on those systems, I can go to that area of a source file and actually see change history, right there in the file. Now that _does_ present its difficulties, too, and SVN does not even have a log tag, so it's not relevant here. However, at least knowing who made the last change and when is _always_ a useful productivity tool for an active project. >From a pragmatic and operational viewpoint, recent industry press comments that about half of all ISP customers are using broadband modems, while the other half are using dialup modems. This means that probably about half of the development team is likely to have a broadband, always-on connection, while the other half has to dial out, wait for arbitration, and then inquire with the SVN repository about history on files when that is needed, then hang up the modem's telephone line. With an author tag, fewer inquiries need to be made, and I'm all for things that simplify a procedure. I personally have _never_ been on a project where people got into territorial staring contests over ownership of _any_ part of the code base. Source code tags were a productivity tool that everyone used to keep track of what changes were made and when. Such tools have always benefitted these projects, and I think they will benefit this one, too. ...snip... > > My preference is to not have them here in Apache Harmony, but that > said, I want to make sure that contributors are recognized for both > general participation as well as significant 'bulk' contributions. > To solve that, I can think of two things offhand : > > 1) We should have a page like the HTTP project (you know, the "Apache > webserver") > > http://httpd.apache.org/contributors/ > > where we have a list of our committers and their ongoing activities, > and a section noting the contributions that the project accepted. > > 2) In order to get attribution closer to the code, we could also have > an "AUTHORS" file per module, so that we'd easily know who is working > on what - if you are a committer working on a module, you'd add your > name to the list. Additionally, if there was a bulk contribution > that seeded a module (like the three contribs we have now), we can > have a note about that at the top of the AUTHORS file such as > "ArchieVM originally contributed by Archie Cobbs" (yeah, I know we > aren't calling it ArchieVM...) or something like that. > > Thoughts? SOAPBOX #2 ---------- I have looked at the HTTPD contributors page and I like it. I have also looked at the AUTHORS file in Archie Cobbs' code and I like it as well. If I had to choose between the two, I'd choose both. Why? Because while the web page is a very visible location, the source code will travel far and wide. Many users will not ever even _consider_ looking at the web site except to locate the download package. All they want is the code so they can compile it and get on with their work. I know. I've been there several times for different packages. Having these attributions bound together with the source code in this way means that that the work and the credits stay together. As a user of such packages, I _do_ read the credits as a part of what I downloaded so I learn everything about a package that I can in order to do my job properly with the task and download module at hand. Dan Lydick > > geir > > > -- > Geir Magnusson Jr +1-203-665-6437 > [EMAIL PROTECTED]
