> On 11 mar 2015, at 20:37, Martin Buchholz <marti...@google.com> wrote: > > Producing good error messages is such hard work!
Aye. And so often forgotten. > > Instead of 0%3o, use 0%03o > Since you want to print the lowest 9 bits of the mode, don't you want & 0x1ff Absolutely. I opted for the octal representation 0777 instead which seemed fitting here. new webrev: http://cr.openjdk.java.net/~sla/8074812/webrev.02/ <http://cr.openjdk.java.net/~sla/8074812/webrev.02/> Thanks, /Staffan > > On Wed, Mar 11, 2015 at 2:30 AM, Staffan Larsen <staffan.lar...@oracle.com > <mailto:staffan.lar...@oracle.com>> wrote: > Thanks for the feedback. Here is a new version that prints out more details > for each of the errors messages. Let me know if you have suggestions for > better wording. It also adds an #include for jvm.h that was missing from some > of the files (it is needed for jio_snprintf). > > webrev: http://cr.openjdk.java.net/~sla/8074812/webrev.01/ > <http://cr.openjdk.java.net/~sla/8074812/webrev.01/> > > /Staffan > >> On 10 mar 2015, at 19:07, Martin Buchholz <marti...@google.com >> <mailto:marti...@google.com>> wrote: >> >> >> >> On Tue, Mar 10, 2015 at 10:53 AM, Jaroslav Bachorik >> <jaroslav.bacho...@oracle.com <mailto:jaroslav.bacho...@oracle.com>> wrote: >> >> This just got me thinking - would including [sb.st_uid, uid] and [sb.st_gid, >> gid] in the error message be of any additional benefit? >> >> Yes. How much do you want to improve the quality of error messages? >> >> You could use the word "effective" only when effective and real users don't >> match. >> You could print out the two mismatched values. > >