Menu
▾
▴
[Tcl-bugs] [ tcl-Bugs-2762041 ] http11.test test case fails
|
From: SourceForge.net <no...@so...> - 2009-09-10 15:59:09
|
Bugs item #2762041, was opened at 2009-04-14 09:49 Message generated for change (Comment added) made by lvirden You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=110894&aid=2762041&group_id=10894 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: 29. http Package Group: development: 8.6b1.1 Status: Open Resolution: None Priority: 5 Private: No Submitted By: Larry W. Virden (lvirden) Assigned to: Pat Thoyts (patthoyts) Summary: http11.test test case fails Initial Comment: I downloaded the latest Tcl cvs head from the activestate ftp site. The following tests in http11.test failed. I last tested at the end of May and don't recall seeing these failures. http11.test ==== http-2.6 -channel,encoding gzip,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding gzip}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close gzip {} 121845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close gzip {} 0 ==== http-2.6 FAILED ==== http-2.7 -channel,encoding deflate,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding deflate}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close deflate {} 121 845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close deflate {} 0 ==== http-2.7 FAILED ==== http-2.8 -channel,encoding compress,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding compress}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close compress {} 12 1845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close compress {} 0 ==== http-2.8 FAILED ---------------------------------------------------------------------- >Comment By: Larry W. Virden (lvirden) Date: 2009-09-10 11:59 Message: Just a note that this continues to be an issue, 5 months later. on Solaris 8. ---------------------------------------------------------------------- Comment By: Larry W. Virden (lvirden) Date: 2009-06-26 10:32 Message: Note that today, June 26, 2009, I downloaded the tcl cvs head from the activestate ftp site, configured and built it on my solaris 8 machine using Sun's C compiler, and here's what I see for these tests: http11.test ==== http11-2.6 -channel,encoding gzip,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding gzip}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close gzip {} 121845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close gzip {} 0 ==== http11-2.6 FAILED ==== http11-2.7 -channel,encoding deflate,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding deflate}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close deflate {} 121 845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close deflate {} 0 ==== http11-2.7 FAILED ==== http11-2.8 -channel,encoding compress,non-chunked FAILED ==== Contents of test case: set tok [http::geturl http://localhost:$httpd_port/testdoc.html?close=1 -ti meout 5000 -channel $chan -headers {accept-encoding compress}] http::wait $tok seek $chan 0 set data [read $chan] list [http::status $tok] [http::code $tok] [check_crc $tok $data] [meta $tok connection] [meta $tok content-encoding] [meta $tok transfer-encoding] [expr {[ file size testdoc.html]-[file size testfile.tmp]}] ---- Result was: ok {HTTP/1.1 200 OK} {crc32 mismatch: c554bc27 ne edc98c73} close compress {} 12 1845 ---- Result should have been (exact matching): ok {HTTP/1.1 200 OK} ok close compress {} 0 ==== http11-2.8 FAILED ---------------------------------------------------------------------- Comment By: Pat Thoyts (patthoyts) Date: 2009-06-25 18:59 Message: It is definately some interaction with the 'zlib push' and the fcopy we use in the http package when the -channel option has been provided. The only tests that fail use both -channel and a compression algorithm which leads to a 'zlib push'. Test http11-2.9 is the same as 2.[678] but doesn't use any compression (hence does not 'zlib push' and it never fails. Recognizing this I've added some puts commands into http::CopyDone and arranged to retry when we see an [eof]. On [eof] I set a flag in the state array and try one more time. This causes the tests to pass now and in the following output we can see that 2.7 had an early [eof] then continued. I suspect that given more data we might see additional [eof] later so a single retry is probably not robust. The channel termination can't be due to the data being sent - first, -eofchar is {} here and secondly the three failing tests use different compression algorithms so the data stream will be different in each case. Tests began at Thu Jun 25 22:03:25 BST 2009 http11.test == CopyDone: count:8192 total: 8192 err:'' == Eof '1' == Retry on Eof == CopyDone: count:8192 total: 16384 err:'' == CopyDone: count:8192 total: 24576 err:'' == CopyDone: count:8192 total: 32768 err:'' == CopyDone: count:8192 total: 40960 err:'' == CopyDone: count:8192 total: 49152 err:'' == CopyDone: count:8192 total: 57344 err:'' == CopyDone: count:8192 total: 65536 err:'' == CopyDone: count:8192 total: 73728 err:'' == CopyDone: count:8192 total: 81920 err:'' == CopyDone: count:8192 total: 90112 err:'' == CopyDone: count:8192 total: 98304 err:'' == CopyDone: count:8192 total: 106496 err:'' == CopyDone: count:8192 total: 114688 err:'' == CopyDone: count:8192 total: 122880 err:'' == CopyDone: count:7157 total: 130037 err:'' == Eof '1' ++++ http11-2.7 PASSED == CopyDone: count:8192 total: 8192 err:'' == CopyDone: count:8192 total: 16384 err:'' == CopyDone: count:8192 total: 24576 err:'' == CopyDone: count:8192 total: 32768 err:'' == CopyDone: count:8192 total: 40960 err:'' == CopyDone: count:8192 total: 49152 err:'' == CopyDone: count:8192 total: 57344 err:'' == CopyDone: count:8192 total: 65536 err:'' == CopyDone: count:8192 total: 73728 err:'' == CopyDone: count:8192 total: 81920 err:'' == CopyDone: count:8192 total: 90112 err:'' == CopyDone: count:8192 total: 98304 err:'' == CopyDone: count:8192 total: 106496 err:'' == CopyDone: count:8192 total: 114688 err:'' == CopyDone: count:8192 total: 122880 err:'' == CopyDone: count:7157 total: 130037 err:'' == Eof '1' == Retry on Eof == CopyDone: count:0 total: 130037 err:'' == Eof '1' ++++ http11-2.9 PASSED ---------------------------------------------------------------------- Comment By: Emiliano (egavilan) Date: 2009-05-11 12:12 Message: I can confirm the same failures on netbsd 5.0, on both 32 bits (i386 port) and 64 bits (amd64 port). I think bug 2780687 is a duplicate if this. ---------------------------------------------------------------------- Comment By: Don Porter (dgp) Date: 2009-05-07 13:52 Message: Just testing today on a 64-bit system: $ uname -a Linux <REDACTED>.nist.gov 2.6.18-128.1.6.el5 #1 SMP Wed Apr 1 09:10:25 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux and I see tests http-2.7 and http-2.8 reliably fail there just as they do in the original report here. No failure of http-2.6 though. ---------------------------------------------------------------------- Comment By: Pat Thoyts (patthoyts) Date: 2009-04-19 19:20 Message: these are new files. For some reason we are having trouble writing to a file via -channel so the crc and the size checks are wrong. Reproduced on my solaris10 box. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=110894&aid=2762041&group_id=10894 |