RE: Bug#276126: [exim] allow headers_remove|add options to b…

Inizio della pagina
Delete this message
Reply to this message
Autore: Eli
Data:  
To: 'Marc Haber', exim-users
CC: 
Vecchi argomenti: Re: Bug#276126: [exim] allow headers_remove|add options to be given multiple times
Oggetto: RE: Bug#276126: [exim] allow headers_remove|add options to be givenmultiple times
Marc wrote:
> On Wed, Feb 16, 2005 at 09:44:52AM +0000, Philip Hazel wrote:
>> OK. I will try to remember that when a Debian person says "bug" they
>> don't necessarily mean it. I won't therefore give reports containing
>> the word "bug" as much priority as I might have. That is, I won't look
>> at them as urgently as I normally do.


> I am pretty well aware that Debian is unpopular with exim upstream for
> shipping ancient versions and for using a non-standard configuration
> scheme since we need to be runnable out of the box. But I don't think
> that it is a good idea to ignore bug reports coming from Debian.
> Remember that we have found real bugs in the past. Please reconsider
> this decision, which I am rather disappointed with.


You mis-interpret his words (not unlike mis-interpreting the meaning of
"bug" wrt software). He says he will not look at "bug" reports from Debian
with as much urgency as he normally would, expecting an actual bug in need
of fixing.

Wishlist items should not be classified as bugs when dealing with software
management. It causes people to think there are errors in the program when
in fact you are looking for a feature extension or enhancement which is in
no way a software error.

Anyways, I totally side with Philip, as I too use the subjects and reports
of bug in this list to check for possible issues that could be security
related and that I as a system admin would need to know about. Ending up
reading through a bunch of wishlist items is not my idea of entertaining
when I get several hundred emails coming through various accounts each day
plus trying to do my actual work.

Now, on the issue of headers_add and headers_remove. It's a nice idea, but
should be postponed until such time that we may rework the entire headers
code for other reasons that have been brought up last year (which Philip is
aware of). If your sole reason for wanting to use this multiple times is
for code modularity, use the previously mentioned changes (by using your
ifdefs inside your headers_add/remove sections), or heck, just write a
different router for each module and do it that way.

Also note you say Debian does use a "non-standard configuration shceme".
Debian must adhere to Exims rules for configuration - not the other way
around. If Exim wants to implement something a certain way, just because
you don't like it, doesn't mean it's wrong :)

Besides, if we all thought like this, I'd be submitting bugs to almost every
software vendor out there telling them all to adhere to strict rules with
regards to formatting configuration files (for unity among formats), and
directory structures, use of autoconf (hint hint), etc... - but I don't
because I'm not the boss, I'm just a user.

Wishlist requests aren't evil though, don't get me wrong - just don't expect
everything you ask for when not enough people ask for the same thing. When
it comes to getting ideas to conception, the more people making the same
request pushes development.

Eli.