<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <br>
    <br>
    <div class="moz-cite-prefix">On 1/14/19 7:38 PM, Liam Miller-Cushon
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAL4QsgvcS94rJktT9NJBcXmg-xnp3LZm5iAAw2z6U6GOoDj__A@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div>My concern was mostly that users might forget to include
          the separate file in their report, and by the time it's
          triaged it's probably too late to get that information. <br>
        </div>
      </div>
    </blockquote>
    <br>
    there is an early triage that could ask users for the file /
    parameters plus the release file which could also be helpful to
    reproduce the issue<br>
    <br>
    <blockquote type="cite"
cite="mid:CAL4QsgvcS94rJktT9NJBcXmg-xnp3LZm5iAAw2z6U6GOoDj__A@mail.gmail.com">
      <div dir="ltr">
        <div><br>
        </div>
        <div>Making it easy to copy-paste the arguments as part of the
          crash message could help, but I understand the concern about
          console spam. If you were to print a subset, the non-file
          options are probably the most interesting (especially
          non-standard options, as in e.g. JDK-8059453).</div>
        <div><br>
        </div>
        <div>But having the full list of arguments is even better. Maybe
          add a reminder to the diagnostic text to include the params
          file when reporting a bug (possibly in the "Include your
          program and the following diagnostic" part of msg.bug)?</div>
      </div>
    </blockquote>
    <br>
    I have modified the msg.bug to mention javac parameters too, see
    [1], this version also prints the parameters to the output in case
    of error<br>
    <br>
    [1] <a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~vromero/8216529/webrev.02/">http://cr.openjdk.java.net/~vromero/8216529/webrev.02/</a><br>
    <br>
    Thanks,<br>
    Vicente<br>
    <br>
    <blockquote type="cite"
cite="mid:CAL4QsgvcS94rJktT9NJBcXmg-xnp3LZm5iAAw2z6U6GOoDj__A@mail.gmail.com"><br>
      <div class="gmail_quote">
        <div dir="ltr">On Mon, Jan 14, 2019 at 4:05 PM Jonathan Gibbons
          <<a href="mailto:jonathan.gibbons@oracle.com"
            moz-do-not-send="true">jonathan.gibbons@oracle.com</a>>
          wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
          0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
          <div bgcolor="#FFFFFF">
            <p><br>
            </p>
            <br>
            <div class="gmail-m_8259471553016102052moz-cite-prefix">On
              01/14/2019 02:59 PM, Vicente Romero wrote:<br>
            </div>
            <blockquote type="cite"> <br>
              <br>
              <div class="gmail-m_8259471553016102052moz-cite-prefix">On
                1/14/19 4:20 PM, Jonathan Gibbons wrote:<br>
              </div>
              <blockquote type="cite"> Do you mean the trimming done by
                jtreg?   That will obviously only happen if the crash
                happens in the context of a test.<br>
              </blockquote>
              <br>
              no I saw that issue for bugs breaking the JDK build and I
              saw that the output was incomplete<br>
            </blockquote>
            <br>
            It's still the case that what you describe is our internal
            infrastructure trimming the output, and most users will not
            suffer the same restrictions. <br>
            <br>
            <blockquote type="cite"> <br>
              <blockquote type="cite"> <br>
                As a general solution, if you can't write the file, you
                could write the non-file options, and the first N files,
                to the console, for some suitable N. That being said, if
                you're just printing argv, you don't easily have the
                analysis of file vs non-file options, so it may just
                come down to print the first N args, for a different,
                bigger N.<br>
              </blockquote>
              <br>
              I tend to believe that the general case in which this will
              be most useful will be cases with a large number of
              parameters, so I tend to think that it should be better to
              generate no output rather than generate an incomplete
              output. But I could be convinced otherwise :)<br>
            </blockquote>
            <br>
            Maybe. "No output" means no way of guessing what the input
            was;  "Some output" means you might at least show enough to
            show all the options, even if you don't show all the files. 
            But either way, if you fail to write the output, you should
            report that to the console, so that folk who are expecting
            to see the output and not surprised when it is not
            generated.<br>
            <br>
            <blockquote type="cite"> <br>
              <blockquote type="cite"> <br>
                -- Jon<br>
              </blockquote>
              <br>
              Vicente<br>
              <br>
              <blockquote type="cite"> <br>
                <br>
                <br>
                <div class="gmail-m_8259471553016102052moz-cite-prefix">On
                  01/14/2019 12:03 PM, Vicente Romero wrote:<br>
                </div>
                <blockquote type="cite"> Hi Liam,<br>
                  <br>
                  <div
                    class="gmail-m_8259471553016102052moz-cite-prefix">On
                    1/14/19 2:38 PM, Jonathan Gibbons wrote:<br>
                  </div>
                  <blockquote type="cite">
                    <p>The intent is the crash report should contain the
                      path of the file, and the file should be in
                      "@-file" format, to make it easier to rerun the
                      compilation using the file with `javac
                      @javac.<date>.args`.   If an error occurs
                      when writing to the file, the info could be
                      written to the console, but note that there could
                      be a huge amount of output, depending how many
                      files are specified on the command-line.</p>
                  </blockquote>
                  <br>
                  right, what Jon says. I decided not to write to the
                  console because if the output is huge, most of it will
                  be lost as it will be trimmed thus reducing the
                  usefulness of it.<br>
                  <br>
                  <blockquote type="cite">
                    <p>-- Jon<br>
                    </p>
                  </blockquote>
                  <br>
                  Vicente<br>
                  <br>
                  <blockquote type="cite">
                    <p> </p>
                    <div
                      class="gmail-m_8259471553016102052moz-cite-prefix">On
                      1/14/19 11:22 AM, Liam Miller-Cushon wrote:<br>
                    </div>
                    <blockquote type="cite">
                      <div dir="ltr">Hi Vicente,
                        <div><br>
                        </div>
                        <div>This looks useful.</div>
                        <div><br>
                        </div>
                        <div>Did you consider printing the args to
                          stderr instead of writing them to a file,
                          maybe as part of msg.bug? If I'm reading the
                          patch correctly it currently uses a file
                          `javac.<date>.args` in the directory
                          javac is running from, which users won't
                          necessarily know about to include in crash
                          reports, and which won't succeed if javac
                          doesn't have permissions to create files in
                          the working directory.</div>
                      </div>
                      <br>
                      <div class="gmail_quote">
                        <div dir="ltr">On Mon, Jan 14, 2019 at 10:51 AM
                          Vicente Romero <<a
                            href="mailto:vicente.romero@oracle.com"
                            target="_blank" moz-do-not-send="true">vicente.romero@oracle.com</a>>
                          wrote:<br>
                        </div>
                        <blockquote class="gmail_quote"
                          style="margin:0px 0px 0px
                          0.8ex;border-left:1px solid
                          rgb(204,204,204);padding-left:1ex">Please
                          review the fix for enhancement [1] at [2]. The
                          idea of this <br>
                          enhancement is to print out to a file the
                          arguments passed to javac is <br>
                          there is a fatal error or if a hidden option
                          is passed to the compiler. <br>
                          This could be very helpful to reproduce some
                          bugs reported by users.<br>
                          <br>
                          Thanks,<br>
                          Vicente<br>
                          <br>
                          PS, thanks to Jon for offline feedback an
                          suggestions<br>
                          <br>
                          [1] <a
                            href="https://bugs.openjdk.java.net/browse/JDK-8216529"
                            rel="noreferrer" target="_blank"
                            moz-do-not-send="true">https://bugs.openjdk.java.net/browse/JDK-8216529</a><br>
                          [2] <a
                            href="http://cr.openjdk.java.net/%7Evromero/8216529/webrev.00/"
                            rel="noreferrer" target="_blank"
                            moz-do-not-send="true">http://cr.openjdk.java.net/~vromero/8216529/webrev.00/</a><br>
                        </blockquote>
                      </div>
                    </blockquote>
                  </blockquote>
                  <br>
                </blockquote>
                <br>
              </blockquote>
              <br>
            </blockquote>
            <br>
          </div>
        </blockquote>
      </div>
    </blockquote>
    <br>
  </body>
</html>