| Submitter | Jun Wu |
|---|---|
| Date | April 10, 2016, 12:42 a.m. |
| Message ID | <3efb9e44023397f21243.1460248979@x1c> |
| Download | mbox | patch |
| Permalink | /patch/14478/ |
| State | Changes Requested |
| Delegated to: | Yuya Nishihara |
| Headers | show |
Comments
Jun Wu wrote: > backtrace. We have ignored KeyboardInterrupt already, and it makes It makes > sense to ignore SystemExit as well. Unfortunately this sentence doesn't make sense :( probably drop "makes It makes"
On 04/10/2016 03:14 AM, timeless wrote: > Unfortunately this sentence doesn't make sense :( > probably drop "makes It makes" Sorry about these two patches. :( There must be something wrong with my editor.
On Sun, 10 Apr 2016 01:42:59 +0100, Jun Wu wrote: > # HG changeset patch > # User Jun Wu <quark@fb.com> > # Date 1460248929 -3600 > # Sun Apr 10 01:42:09 2016 +0100 > # Node ID 3efb9e44023397f212437d89bfd7268cb5a15df6 > # Parent a212504cf958742becab6fe518dab9b838f5bbf4 > chgserver: do not print traceback on SystemExit > > Before the patch, if some extension uses "sys.exit", chg will print an extra > backtrace. We have ignored KeyboardInterrupt already, and it makes It makes > sense to ignore SystemExit as well. Can you show the stacktrace? I'm interested in which command can execute that ugly sys.exit(). For "runcommand", SystemExit should be caught at dispatch. > diff --git a/hgext/chgserver.py b/hgext/chgserver.py > --- a/hgext/chgserver.py > +++ b/hgext/chgserver.py > @@ -559,7 +559,7 @@ > except IOError as inst: > if inst.errno != errno.EPIPE: > raise > - except KeyboardInterrupt: > + except (KeyboardInterrupt, SystemExit): > pass If it is a common problem, this line should be copied to commandserver.py.
On 04/10/2016 12:36 PM, Yuya Nishihara wrote: > Can you show the stacktrace? I'm interested in which command can execute > that ugly sys.exit(). For "runcommand", SystemExit should be caught at > dispatch. This was a little hack at facebook that suppresses the error message otherwise would be printed by dispatch.py: https://bitbucket.org/facebook/hg-experimental/src/a5aaa/errorredirect.py#errorredirect.py-48 > If it is a common problem, this line should be copied to commandserver.py. Good point. I will do that in V2.
On Sun, 10 Apr 2016 13:29:58 +0100, Jun Wu wrote: > On 04/10/2016 12:36 PM, Yuya Nishihara wrote: > > Can you show the stacktrace? I'm interested in which command can execute > > that ugly sys.exit(). For "runcommand", SystemExit should be caught at > > dispatch. > > This was a little hack at facebook that suppresses the error message otherwise > would be printed by dispatch.py: > > https://bitbucket.org/facebook/hg-experimental/src/a5aaa/errorredirect.py#errorredirect.py-48 I guess it will be changed to hook dispatch.handlecommandexception(), which was recently added. And hopefully we won't need V2. > > If it is a common problem, this line should be copied to commandserver.py. > > Good point. I will do that in V2.
On 04/10/2016 02:54 PM, Yuya Nishihara wrote: > I guess it will be changed to hook dispatch.handlecommandexception(), which > was recently added. And hopefully we won't need V2. Yep. That's also the reason why it was added. I will graft related commits and make things work (I'm doing the release internally).
Patch
diff --git a/hgext/chgserver.py b/hgext/chgserver.py --- a/hgext/chgserver.py +++ b/hgext/chgserver.py @@ -559,7 +559,7 @@ except IOError as inst: if inst.errno != errno.EPIPE: raise - except KeyboardInterrupt: + except (KeyboardInterrupt, SystemExit): pass finally: sv.cleanup()