Repository navigation
Claude/upgrade ape c library mm z gd - #604
Merged
Merged
Conversation
… identical Re-run under cmd/tclsh/tcltest, which `mk install' now builds: baseline: Total 68118 Passed 62138 Skipped 5916 Failed 64 with -J: Total 68118 Passed 62156 Skipped 5916 Failed 46 Total and Skipped identical to the digit, so nothing became unrunnable and no file aborted -- Passed +18 and Failed -18 is one statement about the same 68118 executions. socket_inet.test has left the failing-file list, and sockets are half the measured blast radius (sockaddr_in 24 -> 16). The refutation condition written before the switch -- a NEW socket failure -- did not fire. The pthread half is NOT measured: 197 thread and 12 testmutex are skipped, so pthread_mutex_t 56 -> 40 was barely exercised. Recorded rather than glossed. Still to check per name: the totals cannot tell 18 fixed from 20 fixed and 2 broken, and socket_inet-5.1/5.3 were recorded as tests that should fail -- if they are among the eighteen that is the leftover-listener false pass, not a fix. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WGAwvvTwDg2yknFkmZ3qzs
… on the VM cfront rebuilt with the old layout rule faults again at pc=0x280db8, the same address to the digit, with cfront 1265 bytes -- byte for byte the pre-J run. The sualign fix is in both builds, since that line was never gated on conformalign, so it is excluded and -J is the vehicle with nothing left over. The pair also hands over a finding: table.c carries seven `if ((long long)...string < 0x200000)' guards, this fork papering over a garbage const char * rather than diagnosing it. They catch only the case where the misread bytes are a small number, and the one that crashed held 0x6f69736963657270 -- the text `precisio' -- far above the threshold. They never worked. Verdict on cfront recorded: the stated reason for keeping it has expired, because the host reproduces the remaining diagnostics character for character, so what is left is upstream's 64-bit unfitness rather than a kencc or libap bug. Keep the directory and stop spending rounds on it: the -9/-J pair is the only end-to-end regression instrument for the layout rule, cfrontsz-probe reads cfront's generated C as its size oracle, and the EDG path would want the same driver and headers. c++lib stays commented out. dupbuf-test on the VM: _dupmark = 1, all seven sections PASS, 0 failures. 4b takes F_DUPFD >= 25 where it used to refuse, and 5 reads the dup without EBADF. The errno 3 there is EWOULDBLOCK from the test's own non-blocking reads, which is the recorded copy-process limit rather than a defect; section 6 reads all eight bytes from the original. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WGAwvvTwDg2yknFkmZ3qzs
table.c papered over a garbage `const char *' in seven places with `if ((long long)...string < 0x200000) continue;'. They never worked: they catch only the case where the misread eight bytes happen to be a small number, and the pointer that actually crashed held 0x6f69736963657270 -- the text `precisio' -- far above the threshold. With -J restoring `name' to 144 the misread is gone and the guards are dead weight that would mask a recurrence. Five were a standalone goto nxt/continue; look__6ktable's was half of an &&; insert__6ktable's was a two-line ||. gcc compiles the file before and after, which is a syntax check and not a control, since both sides pass. The VM is what confirms it: `mk install' in cmd/c++lib should print exactly what it prints now, cfront 5747 with the four iostream.h redeclarations and `two definitions of norm()'. Anything else means a guard was load-bearing, which is a finding rather than a regression. Also recorded: the -J rebuild was run a second time and reproduced cfront 5747 and the same diagnostics, so neither side of the -9/-J pair was a one-off. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WGAwvvTwDg2yknFkmZ3qzs
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
2 participants
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.