Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins that include changes to files matching 'src/*'
|
2026-02-01
| ||
| 21:50 | Fix for hang problem in invert-deposit from claude.ai leaf check-in: ed533e67c3 user: pi tags: invert-deposit | |
|
2025-11-09
| ||
| 04:58 | Pull in fixes from Trunk. check-in: 1e1e7f934b user: poetnerd tags: invert-deposit | |
|
2025-11-03
| ||
| 06:00 | Fix for v3f no-scope rubout support: 1. Decided to patch the version of SET inherited from v3d. Simplest to add the patch script, and do it with a one-liner in scripts/os8/all-tu56.os8. 2. Updated LCSYS.BI and UCSYS.BI as I did for v3d and OCK. check-in: ee7647f03e user: poetnerd tags: trunk | |
| 04:00 | Pull in updates from trunk. check-in: f20057b1c1 user: poetnerd tags: invert-deposit | |
| 03:40 | Updated UCSYS.BI and LCSYS.BI for v3d and ock. TODO: Consider fixing v3f as well. check-in: 519b95ed4b user: poetnerd tags: trunk | |
| 02:46 | Fix src/ock/CUSPS/SET.PA to use correct address for page zero literal for backslash. This gets the stock build of the OS/8 Combined Kit properly supporting SET TTY NO SCOPE. Fixes to LCSYS.BI and UCSYS.BI are also needed because they too break printing rubout support. check-in: ac04094c0b user: poetnerd tags: trunk | |
|
2025-10-25
| ||
| 21:43 | Branch to test software support of reversing the sense of the Deposit switch. This enables changing the switch orientation to match real the PDP-8 without having to cut traces. check-in: 47f170326c user: poetnerd tags: invert-deposit | |
| 16:36 | Update of palbart from Dave Gesswein. Changes: /* v2.14 07Mar23 DJG Fixed compile warning about possible buffer overrun.*/ /* v2.15 31Oct23 FPW Fixed FLTG, patch from Folke Will. */ /* FLTG is palbart extension from MACRO-8 */ /* 02Jan24 DJG Added -b to allow code PAL8 accepts but bad practice*/ /* v2.16 28May24 DBA Added RK8E instructions */ Testing: Output of current .lst files generated by make is the same. Answers ticket 0e69c913aaaa7b13c2a9be95003c12a0268578fc check-in: cb831fcc7a user: poetnerd tags: trunk | |
|
2025-10-16
| ||
| 14:17 | Bring in U/W FOCAL replacement module for operation without EAE hardware. Document the work. Add test code. Add stand-alone test script. Update uwfocal-rk05.os8 to build UWFNOE.SV and UWFTST.SV execuatables. Also update Changelog.md check-in: 1e998eb57e user: poetnerd tags: trunk | |
|
2025-10-12
| ||
| 14:37 | Add FPTEST.TX, a text-based Floating point test suite. check-in: c3cd90243f user: poetnerd tags: uwf | |
| 00:00 | A little more cleanup: Renamed TESTME to SATEST for "Stand-Alone Test". Pulled setting of In-line emulation testing under U/W FOCAL under the control of 8NT.PA which consists solely of EMTEST=1. Updated uwfocal-rk05.os8 script to create 3 versions of U/W FOCAL: UWF16K the EAE-only Legacy system. UWFNOE that contains the emulator. UWFTST that contains the emulator with the in-line testing enabled. check-in: c3e4881678 user: poetnerd tags: uwf | |
|
2025-10-11
| ||
| 02:55 | Give proper credit to Claude and to DECUS 8-436. check-in: 03999ddd60 user: poetnerd tags: uwf | |
| 02:50 | Found root cause of unsuccesful transcendental testing, the transcendentals begin with F. So it's "FATN" not "ATN". check-in: 515cdbc6cf user: poetnerd tags: uwf | |
| 02:37 | Figured out how to add XLIST to hide the stand-alone test suite when we build 8NFIO as a module in U/W FOCAL. check-in: 1ca49c34bc user: poetnerd tags: uwf | |
| 01:54 | Update the documentation file NONEAE.md to reflect current reality. TODO: Find out why transcental functions all return 0.000000000E+00 for both EAE and NON-EAE U/W FOCAL. check-in: 9955158285 user: poetnerd tags: uwf | |
|
2025-10-07
| ||
| 03:07 | Smaller Multiply/Divide enables us to move SWPEM into the page with MUY and DIV, eliminating use of the possibly precious memory elsewhere. check-in: a938e79032 user: poetnerd tags: uwf | |
| 02:54 | I merged the stand-alone and in-line emulation testing. It may be a little slower to tease out root causes, but duplicated code is gone. check-in: a31eaaa143 user: poetnerd tags: uwf | |
| 02:04 | Integrated the division algorithm from DECUS 8-436. They missed an optimization that I made. Resulting code is much smaller. I compared my multiply emulation against DECUS 8-436. I believe mine is smaller and more accurate. All current tests pass. check-in: 2e63f39cf0 user: poetnerd tags: uwf | |
|
2025-10-06
| ||
| 21:14 | Fixed the problem with overflow detection aided by DECUS 8-436 and Claude. This now passes all known tests. check-in: fe92fd8b3e user: poetnerd tags: uwf | |
| 16:25 | A little tightening of code so it will integrate into U/W FOCAL. New failure case, dividend =0 should return AC=0 MQ=0 but currently complains of underflow. check-in: 853df31a5c user: poetnerd tags: uwf | |
| 03:31 | Reverting a bunch of "cleanups" that made a mess of things. Added a test case 24 that was failing. Root cause: No detection that the remainder became zero, and no more subtraction was needed. However the code seems sub-optimal. check-in: 53d0bf515c user: poetnerd tags: uwf | |
|
2025-10-02
| ||
| 22:39 | Start of cleanup of code. Baseline before we merge EMTEST into TESTME. check-in: ce89df8add user: poetnerd tags: uwf | |
| 04:14 | This version appears to work!!! It passed simple smoke tests in U/W FOCAL: T 3; T PI; T 10 / 3. Also all the hard-coded tests pass. TODO: More testing, and maybe some optimization. check-in: 3cfebe4fab user: poetnerd tags: uwf | |
|
2025-09-30
| ||
| 02:10 | Found and fixed problem with the run-time tester corrupting itself. (Fast exit was leaving args from previous calls unset.) Multiply seems fully working. New test case shows divide is not working yet. Updated fossil-settings/crlf-glob to be quiet about various PDP-8 source trees having cr/lf line endings. check-in: 3328f43127 user: poetnerd tags: uwf | |
|
2025-09-26
| ||
| 20:58 | Yet another snapshot. If I disable the in-line testing multiply emulation seems to work in U/W FOCAL. But with testing enabled, a test case that passes stand-alone chokes. I think I've got some kind of uninitialized data or reuse bug I'm not seeing. Cleaned up a comment in 8NFPP.PA. Fixed how in-line testing was returning EAE rather than Emulated results. Added another test case. check-in: 0fd82c1c85 user: poetnerd tags: uwf | |
|
2025-09-21
| ||
| 16:58 | Renamed and reordered placement of temporaries to make sorting out bugs easier. Tests pass, but UWFNOE still fails in a non-obvious way. So this is a snapshot checkin. check-in: 0aeb272361 user: poetnerd tags: uwf | |
|
2025-09-14
| ||
| 23:26 | Found a test case that fails. 8NFIO will need surgery to fix. Checking in the current broken version as a fall back. check-in: acd3fc43a2 user: poetnerd tags: uwf | |
|
2025-09-13
| ||
| 03:58 | Snapshot: Small update to doc. 8NFIO.PA now auto-detects whether it's stand-alone or part of UWF. A couple bug fixes. New test harness: a table of executions, and a stop with easy display of failure in the AC. It passes all tests in the harness but still doesn't work in UWF. Latest multiply fix: We could enter multiply with a non-zero link that the EAE hardware ignored, but that we used. Maybe a similar bug is in the DVI routine. check-in: b75033af80 user: poetnerd tags: uwf | |
|
2025-09-12
| ||
| 03:47 | Snapshot. Publish current draft of non-EAE support for U/W FOCAL: Merged in trunk changes that came in while I was fiddling with this. Add documentation of what's being done in src/uwfocal/NONEAE.md. Point to that documentation in src/uwfocal/README.md Add 8NFIO.PA and 8NFPP.PA, our non-EAE modules. Update scripts/misc/uwfocal-rk05.os8 to build UWFNOE in addition to UWF16K. Now generate listing and bitmap files for when we will need them. Create new scripts/misc/uwf-noneae-test.os8 to be a quick build of 8NFIO for testing. Current status: The emulation is integrated into U/W FOCAL, and it seems to run without totally killing the interpreter. The EAE emulations pass basic tests. Alas, there are corner cases in the code that are still wrong. Even the simple FOCAL statement, "T 3" does not print a 3.0000000e00 as expected. Instead it leaves the FLAC in a state where no more FPP operations are possible. At least "T PI" produces "3.141721908E+00*" as expected, and it didn't do that yesterday. check-in: 7ada921d22 user: poetnerd tags: uwf | |
|
2025-09-04
| ||
| 21:13 | Pull in cleanups learned when trying to submit pdp8_cpu.c upstream to OpenSIMH. check-in: 97bf475e42 user: poetnerd tags: trunk | |
|
2025-07-06
| ||
| 20:41 | Merge in cycle-accurate support! check-in: a08e131416 user: pi tags: trunk | |
| 04:26 | Cosmetic changes: Remove whitespace that conflicts with upstream. Add JMP/JMS comments from trunk for additional clarity. leaf check-in: 8b2c984e88 user: poetnerd tags: pi5-ils2-bworm-cyclerealistic | |
|
2025-07-05
| ||
| 20:49 | Pulled in current epoch of OpenSIMH (2025-07-05) check-in: 16dcc6c24b user: poetnerd tags: trunk | |
|
2025-06-26
| ||
| 04:18 | Missed a semicolon that only shows up when PIDP8I is defined. OOPS. Fixed. closed check-in: 6554aa2384 user: poetnerd tags: bills-slower-version | |
|
2025-06-21
| ||
| 19:59 | Draft: Attempt to speed up cycle-realistic operation by reverting most of the decode logic. I believe some additional overhead was introduced when cycle-accurate operation rewrote the instruction execution loop. This delta keeps as much of the old instruction decode logic as I could, but does add in the overhead for one extra switch on Major_State, and on instructions that utilize Defer and Execute states. Not extensively tested, but it ran well enough to boot OS/8 so I'm committing it to the tree! check-in: c9676ea6c5 user: poetnerd tags: bills-slower-version | |
|
2024-12-27
| ||
| 02:39 | Fix arguments to and error reporting from map_gpio_for_pidp8i check-in: 3fee766e68 user: poetnerd tags: trunk | |
|
2024-11-10
| ||
| 18:35 | Pull pi5 and ils2 branches into trunk. This enables pi4 and pi5 hardware and improves PiDP-8/i displays. check-in: 0ad7bdeb6b user: poetnerd tags: trunk | |
|
2024-08-03
| ||
| 17:17 | Move declaration of op_code outside #ifdef PIDP8I so it works both with and without gpio. check-in: 9de1597bf4 user: poetnerd tags: pi5-ils2-bworm-cyclerealistic | |
|
2024-06-30
| ||
| 18:59 | Enable ILS mode in single-step/single-instruction mode. Enable ILS in more of the throttling modes, including throttling to very slow speeds like 1 instruction per second check-in: 3b3fcb8834 user: HBEggenstein tags: pi5-ils2-bworm-cyclerealistic | |
|
2024-06-20
| ||
| 00:55 | fixed ILS tweaking check-in: 74efaad4a1 user: HBEggenstein tags: pi5-ils2-bookworm | |
| 00:24 | fixed ILS tweaking check-in: b0f1224117 user: HBEggenstein tags: pi5-ils2-bworm-cyclerealistic | |
| 00:21 | fixed ILS tweaking check-in: 7b450b94dc user: HBEggenstein tags: pi5-ils2-bworm-cyclerealistic | |
|
2024-06-15
| ||
| 21:58 | merge Steve Tockey's branch cycle-realistic into the pi5-ils2-bookworm branch check-in: d413ed639f user: HBEggenstein tags: pi5-ils2-bworm-cyclerealistic | |
|
2024-06-14
| ||
| 23:42 | revert pinctrl chaching optim., not worth it. Moved ILS/NLS switching logic from compile time to runtime / configuration file check-in: ccedb853c9 user: HBEggenstein tags: pi5-ils2-bookworm | |
|
2024-06-09
| ||
| 21:02 | Fix for scanswitch as per email on 5/19. Otherwise pidp8i won't start from systemd. check-in: 32b450529e user: poetnerd tags: pi5 | |
| 20:59 | Fix for scanswitch as per email on 5/19. Otherwise pidp8i won't start from systemd. check-in: 3a1f169d7d user: poetnerd tags: pi5-ils2-bookworm | |
|
2024-06-06
| ||
| 19:54 | performance optimizations check-in: 1aefd0c723 user: HBEggenstein tags: pi5-ils2-bookworm | |
| 00:55 | bug fix for LED update in stopped state, and fixing a potential sign/overflow problem in the dithering of the LED update from the decode loop check-in: d2afc15e23 user: HBEggenstein tags: pi5-ils2-bookworm | |
|
2024-05-31
| ||
| 17:16 | forgotten line from previous checkin check-in: 53d0aa19bd user: HBEggenstein tags: pi5-ils2-bookworm | |
| 17:07 | a)As ILS redraw loop is now executing faster and to accomodate fater Pi models, thefrequency at which LED status is updated is increased, so each redraw has at least 32, better ca 60 sampling points to work on, to use the fully brightness level range. b)The user tweakable RISING_FACTOR and FALLING_FACTOR values now also go thru self-calibration, making them comparable among different users and hence potentially different Raspberry Pi models. check-in: f0642ec578 user: HBEggenstein tags: pi5-ils2-bookworm | |