Preparing os8-v3f-extenstions branch for merge with trunk
(2) By poetnerd on 2018-09-29 02:46:30 as edited by tangent on 2018-09-29 17:21:29 [link]
Let us cast a wider net for discussions of preparing the os8-v3f-extensions branch for merging it with the trunk.
At a high level this means replacing the mkos8 script with hard-coded calls to python expect with a general purpose script runner called os8-run. Scripts have been written that are drop-in replacements.
There are numerous details that need to be reviewed so that the result is close enough to canon that further development in a branch no longer makes sense.
There's one big change that I'll discuss in another topic: The simh python expect main loop is backwards. Fixing this change will be done in a new branch *after* the merge.
The other details center around:
* Minimal Viable Functionality for os8-run
* The options and defaults for OS/8 v3d system packs
* What, if any OS/8 v3f functionality will be available?
* Any replacement of legacy OS/8 media such as os8.tu56.
* Correcting major aesthetic issues in file naming or location.
* Fixing any major bugs detected in advance of the merge.
The current TODO list is:
1. Make applying OS/8 patches to the default bootable rk05 image unconditional. An unpatched rk05 image will be available along with the distribution tu56 images in the media/os8 subdirectory.
2. Shorten the names of the system packs:
1. os8v3d-src.rk05 becomes v3d-src.rk05
2. os8v3d-bin.rk05 becomes v3d-dist.rk05
3. os8v3d-patched.rk05 becomes v3d.rk05 and is the default OS/8 boot image.
3. Simplify the conditional versioning feature in os8-run. (There was some private discussion about it. I got too clever. I'm gonna cut back what's there right now.)
4. Add the ability of os8-run to print the values of the version, and the filename expansions.
5. Clean up names and locations of files used/built for the v3f stuff.
1. Find simpler names.
2. Agree on the locations of intermediate files: rk05 image with sources; rk05 image with both sources and binaries.
6. Create a bootable tu56 image that replaces the current os8.tu56 which is really a found artifact.
It should initially be similar to the existing os8.tu56 in functionality but with updated v3d content. Thought was given to putting v3f onto a tape there, but that seems like a further step.
7. Fix recently discovered bug in the scripts that create the td12k os8v3f bootable tu56 image.
8. Add more stuff to the v3f bootable tape playpen. Possibly create an rk05 image. But that seems like a project all by itself.
I just realized that the ".mkos8" extension should be shortened to ".os8".
The name of the script runner is now "os8-run".
We've wanted to shorten the extension.
The name is totally descriptive.
I'll change the names.
Items 1, 2, and 3 are now done and checked in.
Progress has been made on Item 5.
Items 6, 7, and 8 have all been done.
Makefile.in now creates quite viable TD12K and TC08 DECtape .tu56 images.
The main line of this work is now essentially done.
More of the V3F sources can be built, tested and when qualified can be shifted onto the v3f tu56 images over time by editing all-tu56.os8 trivially for each file.
With os8-run in such good shape, os8-cp and its use in creating v3f-build.rk05 looks somewhat fragile and somewhat ugly.
Warren, could you look at the build rules for
V3F_MADE_RK05 and suggest alternatives?
Ideally there'd be an OS/8 RK05 image with sources that would be updated when files in
and that would drive creating new binaries, and new images.
I think the primary ugliness is the use of
os8-cp and the
actions.txt manifest file.
Okay, but don't expect anything before the weekend. Work's a mess this week.
(10) By poetnerd on 2018-10-04 15:07:39 in reply to 9
Totally understandable. No problem.