Among the flaky tests, it seems that the Azure Pipeline suffers relatively frequently from t5516 failing with the Visual Studio builds. Essentially, we grep for an error message, but that error message is produced twice, once by a fetch and once by the upload-pack spawned from it, and those error messages are usually interleaved because of MSVC runtime fprintf() idiosyncracies. As a consequence, I have to re-run about half a dozen failed builds a day, which I would like to avoid. My plan is therefore to merge this patch into Git for Windows v2.24.0-rc2. The commit message of this patch is based, in part, on https://github.com/gitgitgadget/git/pull/407. This fixes https://github.com/gitgitgadget/git/issues/240. Changes since v2: * Using write_in_full() instead of xwrite() again (to make sure that the entire message is printed). * When vsnprintf() fails, now we at least print the prefix. * The code to check whether prefix was too long no longer tests an inequality between pointers, but between sizes. Changes since v1: * Changed the oneline to be more accurate (thanks Junio). * Improved the commit message (e.g. talking about the xwrite() function this patch uses, rather than the write_in_full() function used by an earlier iteration, thanks Gábor). * Revamped the actual code to account for insanely long prefixes (thanks for the advice, Junio). Johannes Schindelin (1): vreportf(): avoid relying on stdio buffering usage.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) base-commit: 566a1439f6f56c2171b8853ddbca0ad3f5098770 Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-428%2Fdscho%2Ffix-t5516-flakiness-v3 Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-428/dscho/fix-t5516-flakiness-v3 Pull-Request: https://github.com/gitgitgadget/git/pull/428 Range-diff vs v2: 1: e426627e14 ! 1: fce0894ee4 vreportf(): avoid relying on stdio buffering @@ -16,7 +16,7 @@ Let's avoid this predicament altogether by rendering the entire message, including the prefix and the trailing newline, into the buffer we already have (and which is still fixed size) and then write it out via - `xwrite()`. + `write_in_full()`. We still clip the message to at most 4095 characters. @@ -46,6 +46,7 @@ Git Bash on Windows: in this case, `stderr` is not actually truly unbuffered, and needs the extra help. + Helped-by: Jeff King <peff@xxxxxxxx> Helped-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@xxxxxxxxxxx> Helped-by: SZEDER Gábor <szeder.dev@xxxxxxxxx> Helped-by: Junio C Hamano <gitster@xxxxxxxxx> @@ -64,9 +65,9 @@ - vsnprintf(msg, sizeof(msg), err, params); - for (p = msg; *p; p++) { -+ p = msg + off < pend ? msg + off : pend - 1; ++ p = off < pend - msg ? msg + off : pend - 1; + if (vsnprintf(p, pend - p, err, params) < 0) -+ return; /* vsnprintf() failed, there is nothing we can do */ ++ *p = '\0'; /* vsnprintf() failed, clip at prefix */ + + for (; p != pend - 1 && *p; p++) { if (iscntrl(*p) && *p != '\t' && *p != '\n') @@ -76,7 +77,7 @@ + + *(p++) = '\n'; /* we no longer need a NUL */ + fflush(stderr); -+ xwrite(2, msg, p - msg); ++ write_in_full(2, msg, p - msg); } static NORETURN void usage_builtin(const char *err, va_list params) -- gitgitgadget
