[ZCM] [ZC] 741/ 6 Comment "Better FTP error messages"

[Collector: Zope Bugs, Features, and Patches ...]

Issue #741 Update (Comment) "Better FTP error messages"
 Status Accepted, Zope/feature+solution medium
To followup, visit:
  http://collector.zope.org/Zope/741

==============================================================
= Comment - Entry #6 by ajung on Feb 2, 2003 7:06 am


________________________________________
= Comment - Entry #5 by chrisw on Jan 13, 2003 10:50 am

> = Comment - Entry #4 by d.maurer on Jan 7, 2003 1:41 pm
> 
> > ... works only in Debug mode ...
> I tested it in both debug and non-debug mode.

Does it still return a useful message, rather than the old unhelpful one, in non-debug mode?

> However, because many people are very sensitive to see
> backtraces, it include a backtrace only when Zope runs
> in debug mode

That sounds sensible.

> > ... thouroughly tested? ...
> It is only lightly tested. Especially, it is only tested
> with "xemacs/efs".

Does it need more testing? Would it be okay to include it on the HEAD and see what happens?

> The biggest problem will probably be that the patch can
> generate an arbitrary long line.
> Old protocols (such as FTP) tend to place line length
> limits and old FTP clients may not check for buffer overrun.

Hmmm... isn't that only the case if the traceback is included?
I'd prefer to see a patch with just gave the error type and value in both debug and non-debug mode. Maybe the full traceback could be logged to zLOG?

cheers,

Chris
________________________________________
= Comment - Entry #4 by d.maurer on Jan 7, 2003 1:41 pm

> ... works only in Debug mode ...
I tested it in both debug and non-debug mode.

However, because many people are very sensitive to see
backtraces, it include a backtrace only when Zope runs
in debug mode

> ... thouroughly tested? ...
It is only lightly tested. Especially, it is only tested
with "xemacs/efs".

The biggest problem will probably be that the patch can
generate an arbitrary long line.
Old protocols (such as FTP) tend to place line length
limits and old FTP clients may not check for buffer overrun.

It may therefore be wise, not to include tracebacks
at all or only when explicitly requested (e.g. via
an environment variable).
This, however, will require documentation, too.
________________________________________
= Comment - Entry #3 by d.maurer on Jan 7, 2003 1:37 pm

> ... works only in Debug mode ...
I tested it in both debug and non-debug mode.

However, because many people are very sensitive to see
backtraces, it include a backtrace only when Zope runs
in debug mode

> ... thouroughly tested? ...
It is only lightly tested. Especially, it is only tested
with "xemacs/efs".

The biggest problem will probably be that the patch can
generate an arbitrary long line.
Old protocols (such as FTP) tend to place line length
limits and old FTP clients may not check for buffer overrun.

It may therefore be wise, not to include tracebacks
at all or only when explicitly requested (e.g. via
an environment variable).
This, however, will require documentation, too.
________________________________________
= Accept - Entry #2 by chrisw on Jan 6, 2003 12:05 pm

 Status: Pending => Accepted

 Supporters added: chrisw

How come this only works in Debug mode?
Has it been tested thoroughyl enough for me just to merge it into the HEAD of CVS without worrying about testing it?

________________________________________
= Request - Entry #1 by d.maurer on Dec 21, 2002 10:25 am


Uploaded:  "ftpmessages.pat"
 - http://collector.zope.org/Zope/741/ftpmessages.pat/view
FTP error messages are not very helpful.

The attached patch provides for exception type
and value and (in debug mode) a compact traceback
==============================================================