RE: Default servlet regressions
> From: Christopher Schultz [mailto:ch...@christopherschultz.net] > Subject: Re: Default servlet regressions > It's nice to know that the EBCDIC->ShiftJIS conversion is now reliable. We actually do have customers that use that conversion (but perhaps not in Tomcat). - Chuck THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. smime.p7s Description: S/MIME cryptographic signature
Re: Default servlet regressions
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Mark, On 8/2/17 4:35 PM, Mark Thomas wrote: > On 02/08/17 21:23, Christopher Schultz wrote: >> Mark, >> >> On 7/31/17 12:16 PM, Mark Thomas wrote: >>> On 30/07/17 19:04, Mark Thomas wrote: All, It looks like there is another regression in the Default Servlet. While I think I have a patch for this, I'm currently leaning towards doing the following: - Announcing the 8.5.19 and 9.0.0.M25 releases with a warning about the regression. - Reverting the encoding changes in 8.5.x and earlier. >> >>> This has been done. >> - Fixing the latest regression for 9.0.0.M26. >> >>> This has been done. But... >> >>> Thinking about this overnight, I thought of some other >>> possible edge cases. The result is that I have put together a >>> greatly expanded set of tests - currently 680 - using >>> parametrization. >> >> You're a machine. > > Nah. I just know how to wrote nested for loops ;) It's nice to know that the EBCDIC->ShiftJIS conversion is now reliable. >>> Of these, currently 196 (no that isn't a typo) fail. I'm >>> currently working my way through them since some failures are >>> expected (and I need to mark them as such so the tests pass) >>> but I have found some additional edge cases that need to be >>> handled. >> >> Thanks for all your hard work on this. > > You are welcome. But - to be frank - I broke this in the first > place. I feel I should be the one to sort it out. I'm just sorry it > has taken a handful of releases. It's a better product for it. I just finally slew a timezone-related issue that was caused by a vendor stubbornly refusing to provide time zone information in their JSON-based API calls (a killing offense IMO) except when it IS there (fun with SimpleDateFormat! Thanks, Sun/Oracle, for not providing a way to emit/parse ISO-8601 date strings using SimpleDateFormat until Java 8 and then making the documentation unreadable so you have to write unit tests just to find out how it works) and the server is running in yet another time zone (e.g. UTC). Sometimes stuff just takes a while. What's a few dozen more lines of plumbing code between friends? :/ - -chris - -- "If you ever meet me in person, you'll see I'm not such a miserable bastard as this thread would indicate." -me -BEGIN PGP SIGNATURE- Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJZgj27AAoJEBzwKT+lPKRYdcUP/A1FcUv8LnQpFxyGiZWONidZ IchrEe3pILV5WGIlWA61IBwgmMpQUvOXMuaACpojS+h/NFUQMCzBevoh25m8S/2+ 0pzIoQEyZZrK6HXjy0Eog9qREqGcyquBZGhpuD5td3YJQsv+XJ+y11EoPDLuo+GO tTPLRJ8BIW7jCCWIxL9s8oGybXxkN6/VgPI11UIUu4tLxVs/AYM+j/qe+Sd21Sz/ 6RBMHrl1Ac6hQoIQORkqCLN6wp5T85aKenH7L1Hoiv4rVVWPDVohI+aym+HeyxN8 VzfRFQLH89EHDDMADXR6SV0XBjgS2lq7GK+/g6CY4gETEztcNijJOgsb+2IsUmeX w2E2ZzG9CJDGV45m37hNV2q6FhPX5Ve5wZ9ocHY/AOLwZnI1sxEMEjSy/4cidr13 3cT0wVAmoJcX0LHKrwq698clBjn92yOmr1CFqo2gJqb2YGllRq4HoOTJJ/ALzP8C eqhc5dGE5ewq4y1kEmivOqtT/aDMpJ5CLx+YAGd7M77o+tsHrRoYHDqHr0fBl+Be gzWS2mGe2M560muxc6YduHUSaDViYv+mlL3DTEoOd71DDs1ZJGUuOP1ZSbEeAeYS zJtba1KaJUW8FO2bsW+C+sT0/XVAsNbE2b/MymqUivMLh9F8OQAx81dpSq2olN5L WoyBR+2P8j3Mi2hZHAfZ =j1Ba -END PGP SIGNATURE- - To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
Re: Default servlet regressions
On 02/08/17 21:23, Christopher Schultz wrote: > Mark, > > On 7/31/17 12:16 PM, Mark Thomas wrote: >> On 30/07/17 19:04, Mark Thomas wrote: >>> All, >>> >>> It looks like there is another regression in the Default >>> Servlet. >>> >>> While I think I have a patch for this, I'm currently leaning >>> towards doing the following: >>> >>> - Announcing the 8.5.19 and 9.0.0.M25 releases with a warning >>> about the regression. >>> >>> - Reverting the encoding changes in 8.5.x and earlier. > >> This has been done. > >>> - Fixing the latest regression for 9.0.0.M26. > >> This has been done. But... > >> Thinking about this overnight, I thought of some other possible >> edge cases. The result is that I have put together a greatly >> expanded set of tests - currently 680 - using parametrization. > > You're a machine. Nah. I just know how to wrote nested for loops ;) >> Of these, currently 196 (no that isn't a typo) fail. I'm currently >> working my way through them since some failures are expected (and I >> need to mark them as such so the tests pass) but I have found some >> additional edge cases that need to be handled. > > Thanks for all your hard work on this. You are welcome. But - to be frank - I broke this in the first place. I feel I should be the one to sort it out. I'm just sorry it has taken a handful of releases. Mark - To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
Re: Default servlet regressions
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Mark, On 7/31/17 12:16 PM, Mark Thomas wrote: > On 30/07/17 19:04, Mark Thomas wrote: >> All, >> >> It looks like there is another regression in the Default >> Servlet. >> >> While I think I have a patch for this, I'm currently leaning >> towards doing the following: >> >> - Announcing the 8.5.19 and 9.0.0.M25 releases with a warning >> about the regression. >> >> - Reverting the encoding changes in 8.5.x and earlier. > > This has been done. > >> - Fixing the latest regression for 9.0.0.M26. > > This has been done. But... > > Thinking about this overnight, I thought of some other possible > edge cases. The result is that I have put together a greatly > expanded set of tests - currently 680 - using parametrization. You're a machine. > Of these, currently 196 (no that isn't a typo) fail. I'm currently > working my way through them since some failures are expected (and I > need to mark them as such so the tests pass) but I have found some > additional edge cases that need to be handled. Thanks for all your hard work on this. - -chris -BEGIN PGP SIGNATURE- Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJZgjTNAAoJEBzwKT+lPKRYiA8QAMaaWrmy5rfbqinOzUnEtNr1 FFbLmOakhZCE0HMoPBby7uln7nXzwSfYj9AOSP/BQOHu17KFjXo2irsf51aQIZQX Wy7QnzBhXg8IO2frFKoSh0FVqzyjMEubwqU0jK/end+snLu8lhg/aaiW6/Z2cH7t AlmOD/z4fVtFdsdkyDVGPVH+DufgZTff6RylqIYI2PFTBAe2JCh7zq8Lcr0TFWYd +lByx2+Na7F1KD47kEBuCNu0n8+uW8f+hnfkBj8HFj1PANd1tYaF2OvrLJsFwBCJ NKQbiTFa1D+LmPHa4gZcq89xbaHO+JoTlAtFOetfWIn5RZ2RUwz37FKUfyFGL1T8 1KbxlR/YwDfMFbp867rP95hAxbjLtlddzvh5Da9qhjhlYTinCtd2nme+TecFKtQf PElx3ySgPtJ8DOTaUD6Xxiz5ZjoeI/X+zcXagEPNf7Q933zsXAlWUlkxqQT+S+ck sN+Pmi/bmD7KVpxeuXcROTbIHJCtMYv2rIskrQcINe9pZXt8X83yj6ceCefi4r/B wLii/lMbTyTzqPVpI7HEkftL6X2zG0fM0cxDRvBh6tBZDfNELRX9xlr7vUAABusk tGsIhXK85JeO9iIis8E7UAXZT1MsaO2Y+e5igRW8/3w3whxfbw8jrrMOmq+xAwR6 NGZjdgwy3faOJUW0+oLa =VkGG -END PGP SIGNATURE- - To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
Re: Default servlet regressions
On 31/07/17 17:16, Mark Thomas wrote: > On 30/07/17 19:04, Mark Thomas wrote: >> All, >> >> It looks like there is another regression in the Default Servlet. >> >> While I think I have a patch for this, I'm currently leaning towards >> doing the following: >> >> - Announcing the 8.5.19 and 9.0.0.M25 releases with a warning about the >> regression. Change of plan on this one. I'm not going to announce at all. I'll just drop them from the mirrors. >> - Reverting the encoding changes in 8.5.x and earlier. > > This has been done. > >> - Fixing the latest regression for 9.0.0.M26. > > This has been done. But... > > Thinking about this overnight, I thought of some other possible edge > cases. The result is that I have put together a greatly expanded set of > tests - currently 680 - using parametrization. Of these, currently 196 > (no that isn't a typo) fail. I'm currently working my way through them > since some failures are expected (and I need to mark them as such so the > tests pass) but I have found some additional edge cases that need to be > handled. 2720 unit tests later (parameterized unit tests are great for really increasing the number of tests) 9.0.x looks good. >> - Tagging and releasing 8.5.20 and 9.0.0.M26 early in August (i.e. early >> this coming week). I'm planning to do this shortly. 9.0.x has passed all my pre-tagging tests so I'll be tagging trunk very soon. >> - Take a longer look at the Default servlet code for M27 that will >> include some refactoring and API changes I've been considering while >> working on the encoding issue. Still on the TODO list for M27 Mark - To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
Re: Default servlet regressions
On 30/07/17 19:04, Mark Thomas wrote: > All, > > It looks like there is another regression in the Default Servlet. > > While I think I have a patch for this, I'm currently leaning towards > doing the following: > > - Announcing the 8.5.19 and 9.0.0.M25 releases with a warning about the > regression. > > - Reverting the encoding changes in 8.5.x and earlier. This has been done. > - Fixing the latest regression for 9.0.0.M26. This has been done. But... Thinking about this overnight, I thought of some other possible edge cases. The result is that I have put together a greatly expanded set of tests - currently 680 - using parametrization. Of these, currently 196 (no that isn't a typo) fail. I'm currently working my way through them since some failures are expected (and I need to mark them as such so the tests pass) but I have found some additional edge cases that need to be handled. > - Tagging and releasing 8.5.20 and 9.0.0.M26 early in August (i.e. early > this coming week). > > - Take a longer look at the Default servlet code for M27 that will > include some refactoring and API changes I've been considering while > working on the encoding issue. > > Thoughts, comments, objections etc.? > > Mark > > > - > To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org > For additional commands, e-mail: dev-h...@tomcat.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org