Open opensdh opened 4 years ago
@mmorris: The question is how much "normalization" we want to do: in at least some of these versions, some files have mixed line endings, some have meaningless execute permission, and there are many ancillary files (e.g., *.class
and *.sln
). To what extent do we want this history to ever include a "complete copy", versus providing a meaningful (if ex post facto) history of the files someone would actually want today)?
It's sounds resaonable to say that it should start simply from whatever code dumps we have, bit-for-bit, but there are many files named things like *.java_old
that would certainly be more useful as old versions rather than as part of the "current" tree. If we make those changes, though, should we just make a completely "clean" history?
@opensdh I'm wondering how would we even determine what is the "best" code for a given module. We currently have no way of running anything of this, right? (short of spinning up a windows 2,000 virtual machine and getting the right java runtime, etc.) If we were starting this project from scratch today, what would we use in place of VRML?
I think the java implementations of the physics phenomenon are interesting in and of themselves. I think isolating them from the legacy boilerplate code might be worthwhile, especially if we can get something that will run on a modern software stack.
Then, as a separate but equally important goal, work toward a "best" copy of the project as it stood circa 2005-2009.
For "best" I think we should just use "newest"—except that I might discard the "ancillary files" (even retroactively throughout the history) that we expect them to be nothing better than a distraction henceforth. I don't know if it's what I would pick today, but the Xj3D that was the replacement for Blaxxun/EAI is still (a little bit) alive at the Naval Postgraduate School, so I don't see why WebTOP should be any less functional today than it was 10 years ago.
So I don't see the two versions as being very different: establish a series of commits that update the appropriate files (without ever having *.java_old
) with appropriate (if approximate) dates, and then make whatever new commits are necessary to make it work (again).
So do you have any objection to making the canonical history be only the real source in the format we'd want now (e.g., just line feeds everywhere)?
@opensdh Oh cool I didn't realize Xj3d was still alive. I'm good with everything you said. No objections here.
Hey guys. Hope everyone is doing well! It's been a while!
Hi Brian! We're going to tidy up this codebase a little. Hey, do you know if one of the last webtop devs might have a more recent version of the code? Or do you think this is probably it? (this is the version you sent me some years ago)
My memory is that this version came directly from the copy that Jeremy posted on Sourceforge. I don't remember what the status was at that time and if any work was done after. I'll create an email thread separately with him.
I do remember some copies of this accidentally having ftp passwords, so we should make sure to change those and scrub them from here.
It would be pretty cool to see if everything still works. I'm guessing it won't be too hard assuming xj3d still works.
We were using subversion for quite a while. I'm not sure if that survived anywhere or not.
WebTOP exists in several places: https://github.com/jedavis82/WebTOP has most of the same code. https://sourceforge.net/projects/webtop-optics/ seems to have no repository but has a source archive that is identical to this repository's state except for some metadata. (Both have some files converted to Unix line endings and lack
+x
permissions, which seems like an improvement.)I also have one complete copy of the Java source (not the HTML) from sometime in 2004 and about two dozen older versions of (parts of) single modules, ranging from 1998 (a very small amount of TOP code) to 2005.
Watch this space for updates on a unified history.