Chinh Nguyen

The following patch: * fixes the assertion and non-blocking problem if a non-zero value is used when creating pipe with `pipe_new` * can reduce the mincap to less than DEFAULT_MINCAP...

There is a commit on https://github.com/git-blame/pipe. This returns thread handles so that in `pipe_test.c` you can join the child threads and clean up resources.

Pipe maintains a circular buffer, where `s.begin + elem_size` points to the 1st elem to pop. There is a sentinel that is `elem_size`. `s.bufend` is the end of the buffer....

The fix is simply making sure `s.bufend - s.begin > elem_size` before copying. ```diff --- a/pipe.c +++ b/pipe.c @@ -960,11 +960,14 @@ static inline snapshot_t pop_without_locking(snapshot_t s, assertume(s.begin != s.bufend);...

See also commits in https://github.com/git-blame/pipe

Note this only occurs in `pipe_test.c` and/or if you use the utilities functions. In any case, there is a commit on https://github.com/git-blame/pipe. This returns thread handles so that in `pipe_test.c`...

_ssl-cert_ script calls function **getCertificate** from script _sslcert_ which usually calls **get_ssl_certificate**. This is a wrapper around openssl function that only returns the end/peer cert. Your best chance, without writing...

Seems to occur when `archive_read_open_memory` is used. See https://github.com/libarchive/libarchive/issues/1233#issuecomment-527934294

> ./bsdtar-old -t -f test/Virtual-Printer-1.0.0-9972168.exe bsdtar-old: Error opening archive: Unrecognized archive format ./bsdtar -t -f test/Virtual-Printer-1.0.0-9972168.exe bsdtar: Invalid CAB header bsdtar: Error exit delayed from previous errors.

I encountered this again recently. I think what's tripping up libarchive are windows "installation" `Exes`. They have a CAB file within the binary. It looks like libarchive detects the CAB...