Add mapping for PF keys on 3278-2 with APL keyboard - #4
Open
hanshuebner wants to merge 21 commits into
Open
hanshuebner wants to merge 21 commits into
hanshuebner wants to merge 21 commits into
Conversation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SSL sockets buffer decrypted data internally, but select() only monitors the underlying OS file descriptor. When a TLS record spans multiple recv() calls, the remaining data sits in SSL's internal buffer invisible to select(), causing long delays until the next network event. Check SSLSocket.pending() before falling back to select() so buffered SSL data is processed immediately. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add ä/ö/ü/Ä/Ö/Ü mappings to CHAR_MAP using the GE/APL character plane (0x50-0x54 lowercase, 0x70-0x74 uppercase). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Creates /dev/ttyCoax symlink based on USB serial number. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add --keyboard-language option (us/de) for 3278 keyboards. The DE variant remaps the number row shifted keys, umlauts on ä/ö/ü keys, ß, § and other German-specific key positions. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Display encryption status and hostname on the terminal status line during TN3270 sessions. Add --no-hostname-status option to suppress the hostname display. Pad OEC indicator to full status line width. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add --log-level command line option (debug/info/warning/error) to control logging verbosity. Add debug log messages for alarm queuing and delivery to help diagnose buzzer issues on physical terminals. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
A full-screen EAB write delivered in a single EAB_WRITE_ALTERNATE command (~3832 interleaved bytes) exceeds what some CUT terminals can buffer. The IBM 3192 silently truncates at its buffer limit, and the IRMA coax adapter rejects the frame with a protocol error (no TT/AR response). Both result in a partially-rendered screen and, because the non-display attribute byte for the password field sits near the end of the write, a password field that echoes. Real IBM 3278-2 terminals happen to accept the full write, which is why the issue was not seen earlier. Make prepare_jumbo_write split at 1024 bytes by default for all paths. The 3299 multiplexer and legacy Arduino-Mega firmware paths already used this limit successfully. COAX_JUMBO=ignore still disables splitting for anyone who needs the old behavior. Also stop silently leaving address_counter stale on a failed write in BufferedDisplay._write_range - null it out so the next flush forces a fresh LOAD_ADDRESS_COUNTER. The regen/EAB buffer and dirty set are already preserved because _commit() does not run on exception, so the next flush retries the range. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The IRMA emulator writes something directly to the left of column 17, which collided with the security indicator. Shift it right by 3 columns. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
pySerial waits for ever on a port opened without a timeout. An interface that stops answering therefore leaves the controller in a read that never returns: no keystrokes, no polling, nothing in the log, and a service manager sees a process that is still running. The interface library raises InterfaceTimeout on a read that comes back empty, and with a timeout set that is what happens -- the run ends and a fresh one takes its place. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
A USB serial device ends a transfer that fills its last packet with a zero length one, and the read that follows the port becoming readable returns nothing. pySerial takes that for a disconnected device and raises, so the run ends over an event that means only "no bytes this time" -- which is what the terminal coming back online does to a session: the controller dies and the screen it had just drawn goes with it. The read is retried until the port's own timeout, so a port that has really gone quiet still ends the run through the timeout that says so. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
pySerial gathers bytes for a read in a buffer of its own and raises if the port then reports data and hands over none -- and that buffer goes with the exception. Retrying the read therefore resumes past the bytes it dropped, and the message stream is read out of step from there: responses come back with lengths that do not match and frames that are a byte long. Reading through the port's descriptor keeps every byte that arrives, ends the read at the port's timeout, and leaves the interface library to report the timeout as its own. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
Reading around pySerial cost more than it bought. Swallowing its "port reported data and gave none" left the bytes it had already gathered behind, so the message stream ran out of step and responses came back with lengths that did not match; reading the descriptor instead met the port's non-blocking mode and ended the run on EAGAIN. Neither appeared while the port was simply given a timeout: that ends a run cleanly on a port that has gone quiet, which is what the timeout is for, and what starts the next one. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
A USB serial device ends a transfer that fills its last packet with a zero length one, and the read that follows the port becoming readable returns nothing: about one attach in eight ended there, with pySerial calling it a disconnected device and the session going with it. Only the single byte read is retried, which is where this happens and where pySerial has gathered nothing it could drop. A larger read keeps its own behaviour, so the message stream cannot resume past bytes lost with an exception, and a port that stays quiet still ends the read at the timeout. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
A USB serial device ends a transfer that fills its last packet with a zero length one, so a port that has just reported data hands over none: about one attach in eight ended there, with pySerial calling it a disconnected device and taking the session with it. Retrying pySerial's read only helped where it had gathered nothing to drop -- asked for more than a byte it drops what it has, and the message stream resumes past the hole and reads lengths that do not match. Reading the descriptor keeps what arrives. Nothing to hand over yet, as an empty read or as EAGAIN on the non-blocking descriptor pySerial opens, waits for more until the port's timeout, and a port that stays quiet ends the read there, which the interface library reports as its own timeout. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
Handing back an empty read is read as the end of the stream: the SLIP decoder flushes what it is holding and gives it up as a message, so an answer still arriving comes out truncated and one that has not started comes out empty. Both reach the interface library as a response it cannot make sense of -- "Invalid response message" and "Response message length mismatch" -- rather than as the wait that they are. A read that gathered nothing by the timeout now says so, and the bytes it did gather are handed over as before for the decoder to carry on with. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018WKhPX5USQQuBdi6jM5nc7
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.