Some tests are not getting posted properly

[mathiasbynens]

The <table> on https://jsperf.com/cash-vs-jquery-parents/2 has an empty <tbody>.

Somehow, no tests were inserted into the database for this page:

mysql> select * from tests where pageID = 1303938;
Empty set (0.00 sec)

Any ideas how this happened?

[bd82]

I think I can reproduce this.

• I made a test case: https://jsperf.com/javascript-parsing-libraries-benchmark/1
• I tried editing it to fix a bad URL. https://jsperf.com/javascript-parsing-libraries-benchmark/1/edit
• I get Not all tests inserted after saving the test.
• If I open the new revision e.g: https://jsperf.com/javascript-parsing-libraries-benchmark/2
  There will be no test cases present in it.

[gionkunz]

This seems to happen for me for every created scenario (just tried to edit one I've just created)

[talee]

Reprod via editing a new one just now.

[japzi]

Happens to me as well, I can't edit a test without getting "Not all tests inserted".

[zwacky]

reproducable on my part as well. trying to edit a test with 2 cases and getting "Not all tests inserted".

The POST request (https://jsperf.com/my-test-name/1/edit) returns 400.

[SeanJM]

reproducible for me too https://jsperf.com/startswith-or-regex2-or-slice-or-substr-or-indexof/1/edit

[Jameskmonger]

Reproducible for me too at https://jsperf.com/array-destroy/224 (edited at https://jsperf.com/array-destroy/224/edit)

[MatthewRalston]

I am getting the same issue. First POST is "not all tests inserted" followed by a locking up of the form with "Please review the required fields and save again", where the field is filled out in the HTML.

[dinazil]

Same here: can't edit https://jsperf.com/parentorselector

[kalvindukes]

Can't even add any tests to edits due to missing buttons and the reproducible errors on this issue

[maxbeatty]

We have enough comments on this being reproducible. Submitting a pull request with a fix is the best contribution.

[maxbeatty]

If you're in a position to sponsor jsPerf, please see #291

[jake-macadamian]

Reproduced: https://jsperf.com/jquery-each-vs-underscore-each-vs-for-loops/104

[manico]

Same here: https://jsperf.com/scd-float-rounding/1/edit

[mathiasbynens]

Thanks to @LeZuse and @maxbeatty #311 was just merged and deployed. Does that help?

[cshaa]

Still having problems with http://jsperf.com/rounding-a-number

[maxbeatty]

Will you please provide more detail? It really helps shorten the time needed to fix

[dcodeIO]

Also still having the "Not all tests inserted" issue:

• Logged in via GitHub
• Redirected to GitHub to authorize
• Redirected to jsPerf, need to click the GitHub button again (did it turn gray from blue?)
• Not sure if I'm logged in
• Created a new test (no name, email, website, just title, prep code, setup, 2 named snippets, non-published to keep fiddling around) and saved
• Revision 1 is ok
• Edited it, added 4 new code snippets (didn't edit anything else), saved
• Redirected to recaptcha (is this from cloudflare?) to approve that I'm a human
• Redirected to jsPerf, showing "Not all tests inserted"
• Tried again
• Tried again adding and removing a space to/from each field
• recaptcha again
• Still just "Not all tests inserted"
• Opening Revision 2 manually shows just the 4 new code snippets
• From there on nothing seems to work anymore (edited different things, not sure what exactly):
  • Keeps redirecting me to the recaptcha thing and showing "Not all tests inserted"
  • Tried deactivating uBlock
  • Revision 3 has just one snippet
  • Revision 4 has none
  • ...
• Gave up

Chrome 55.0.2883.87 m, Win 10

[dcodeIO]

The issue apparently is that, when updating tests, the SQL query contains a WHERE pageID=NEWID clause, where NEWID is given as the page id of the new revision, which updates nothing, because the stored pageID is that of the old revision. Hence, only the new tests are inserted for the new revision's page, afterwards resulting in a "Not all tests inserted" error, which matches what I experienced above. Need to look through this a little more for a PR.

[dcodeIO]

Do you know if it is actually possible to UPDATE an existing test, without creating a new revision? Knowing that would help me fix it.

Edit: Nevermind, looks like this isn't super important.

[rjamesnw]

There really needs to be a way to delete a revision by the OP, otherwise it gets cluttered fast with erroneous posts. Perhaps put a time limit on it if necessary, so after say a month it cannot be deleted, or perhaps if another revision exists after, it can't be deleted, etc.

[jonathanarbely]

Doesn't work for me too.
https://jsperf.com/caching-jquery-selectors

[sylido]

Doesn't work for https://jsperf.com/jquery-each-vs-each-vs-for-loop/ as well :(

[maxbeatty]

We have enough comments on this being reproducible. Submitting a pull request with a fix is the best contribution.

[mindplay-dk]

Doesn't work for me either, trying to fork this one.

[arthurlacoste]

Still happening for me.

[manucorporat]

Still a problem! https://jsperf.com/reusable-args-vs-local-vars/3

[grave-dancer]

best website ever... this problem has still not been fixed as of november 2017 and it seems the staff blame cloudfare

[jwschram]

This is still an issue. My edits to a test suite have totally screwed up the history.

https://jsperf.com/document-fragment-vs-innerhtml-vs-looped-appendchild/

Edits 70-73 are garbage. Is this under active investigation?

[alphanull]

+1 - also a problem for me. Had to manually clear the DB and reenter everything ... (running local installation)

[alphanull]

Basically, jsperf is more or less unusable because of this bug, it JUST happened again, forcing me to restart everything and clear the DB. I also made other observations that might contribute:

• I cannot see "my Tests" anymore after logging in the next day
• The test I have created does NOT show up in "latest", only with "popular"
• Despite the failed attempt to edit my test, a new "revision" (or at least another DB entry) is created nonetheless, but not marked as a new revision. I.e. the two entries I see under "latest" are pointing to the same URL.
• But then, all tests in the tests Table still point to the first entry. maybe thats the source of the error?
• Browserscope ID is different with the new entry, maybe it has something to do with that?
• After a while (8 hours according to docs), jsperf forces a new revision. I think this limit shouldn't be there or at least configurable, esp. on local installations

And it should be possible to have only a local login, or none at all for a local installation, but think this is another topic.

Please, fix this bug, this is quite serious and makes working with jsperf almost impossible right now (at least for me)!