Not delivering {bugs,gnats,faq}.html in releases

[Daniel Berlin]

On Thursday, June 26, 2003, at 2:20 PM, Joseph S. Myers wrote:

> On Thu, 26 Jun 2003, Gerald Pfeifer wrote:
>
>> I had a look, and transforming this into something that is really 
>> suitable
>> for someone without web access (there is a lot of "useless"  stuff 
>> there)
>
> s/without web access/without web access to gcc.gnu.org at that time/.  
> I
> observed the following failure mode of web bug submission last night:
>
> * Log in to Bugzilla, prepare bug report.
> * Try to submit, and find that gcc.gnu.org has gone down since the bug
> report form was retrieved.
> * So need to cut and paste bug report into a file for later submission.

Or just leave that window open until it comes back up.
Or get a browser that knows how to save form data for you.

>
> (Mail is more robust at queuing data when some link in the transmission
> chain is down.

Only because there are multiple links in the transmission chain in the 
common case.
With HTTP, there aren't multiple links.

> A point not I think mentioned in the previous discussions
> of why people prefer gccbug (offline usage, using a real text editor, 
> ...).)
>
Except that we get < 5 gccbug reports a month now.
Sorry, but you keep pointing to these people who "prefer gccbug", and i 
can't find them going through the received gccbug mail archives for the 
past year.
Note that most binary distributions don't even bother to include gccbug.
There aren't large numbers of people who make bug reports up offline, 
apparently, given the decreasing usage of gccbug.

In addition, there are quite a few browsers that will let you edit the 
text entry boxes with your "real" editor.

I'm all for comments and followups being submitted by email. I think 
that's perfectly approriate.
However, I just don't think we need an email bug submission tool.

And just to point out, if the numbers stay on the same path, we won't 
actually be getting any gccbug reports by the time 3.4 is released.

>>> Removing documentation previously present in releases of how to use 
>>> part
>>> of those releases would certainly be a regression.
>>
>> Aren't we talking only about the documentation of (Bugzilla) 
>> components
>> here?
>
> I think essentially so.
>
> (One thing that should be done to the _online_ form - which was a 
> problem
> with GNATSweb, and is still a problem with Bugzilla - is pointing out 
> to
> users that they should change the component from the default "ada" if 
> it
> isn't an Ada bug - some manage to miss this.  Or making the default 
> "---"
> and only accepting the form if a non-default value has been entered.)

I could do this if you like.

>
> -- 
> Joseph S. Myers
> jsm28@cam.ac.uk