Conversation
with the target encoding if the conversion fails
…ring add and clone
|
In recent build only 92 tests were running Fix is to enable tests - this is WIP , to enable all tests.. |
HarithaIBM
requested review from
IgorTodorovskiIBM and
MikeFultonDev
as code owners
🤖 Augment PR SummarySummary: Improves z/OS handling for Git’s working-tree encoding/tagging behavior and expands build/runtime dependencies to support iconv-based conversions. Changes:
Technical Notes: Several changes rely on z/OS-specific APIs ( 🤖 Was this summary useful? React with 👍 or 👎 |
![augmentcode[bot]](https://avatars.githubusercontent.com/in/1027498?s=60&v=4){kind=small}
| + line++; | ||
| + col = 1; | ||
| + } else { | ||
| + col++; |
There was a problem hiding this comment.
In find_first_non_ascii(), col is incremented before checking whether src[i] is the offending byte, so the reported column will be off-by-one (e.g., a bad first byte reports col 2). This likely makes the new test that expects col 1 fail and also makes diagnostics misleading.
Severity: medium
🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.
| + TEST_SHELL_PATH = $(SHELL_PATH) | ||
| + SHELL_PATH_FOR_SCRIPTS = /bin/env bash | ||
| + PYTHON_PATH = python | ||
| + PYTHON_PATH = /home/opnzos/local/pyz/bin/python |
There was a problem hiding this comment.
PYTHON_PATH is hard-coded to /home/opnzos/local/pyz/bin/python, which is likely environment-specific and can break builds/tests on other z/OS setups where Python is installed elsewhere.
Severity: medium
🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.
| export ZOPEN_INSTALL_OPTS="" | ||
| #export ZOPEN_CHECK_OPTS="-i test -j\$ZOPEN_NUM_JOBS" | ||
| export ZOPEN_CHECK_OPTS="test -j\$ZOPEN_NUM_JOBS" | ||
| export ZOPEN_CHECK_OPTS="-i test -j\$ZOPEN_NUM_JOBS" |
There was a problem hiding this comment.
ZOPEN_CHECK_OPTS uses make -i test ..., which ignores test failures; this can cause CI/package builds to look successful even when the test suite is failing, making regressions easy to miss.
Severity: medium
🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.
| + __chgfdccsid(lk->tempfile->fd, 819); | ||
| + struct stat st; | ||
| + if (lk->tempfile && fstat(lk->tempfile->fd, &st) >= 0 && S_ISREG(st.st_mode)) { | ||
| + if (flags & LOCK_TAG_TEXT) |
There was a problem hiding this comment.
On z/OS the lockfile fd is tagged (text) or set binary, but auto-conversion isn’t disabled here; that’s inconsistent with most other z/OS fd handling in this PR and could allow unintended codepage conversion while writing lockfile contents.
Severity: medium
🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.
1 participant
with the target encoding if the conversion fails