Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Fri, 06 May 2005 00:27:44 +0100, Tony Hoyle <tmh at nodomain.org>
wrote:
>Glen Starrett wrote:
>
>> Yikes! I'm going to have to rework my loginfo strategy now :-( . The
>> docs have always said that -- is this new to 2.5.01.*? I have mine set
>> up to progressively get more general, so the correct group gets the
>> correct messages, but if this isn't working in order and stopping at the
>> first then it's going to get hairy.
>
>I've aleays expected it to work that way. It's generally possible to
>make regexps that handle things that way, but it's not possible to
>suddenly make things match multiple times - ie. to work the way the docs
>say at the moment I'd have to remove potentially useful functionality.
Why is this so difficult?
I would assume that reading a file and parsing it line by line would
be very simple:
while no direct match do
{
read line from file
look for direct match
look for DEFAULT
look for ALL
}
if file exhausted then
{
use DEFAULT if found
use ALL if found
}
else if direct match found then
{
use direct match
if ALL found then
{
use ALL entry
}
else while file not exhausted do
{
read line from file
look for ALL
if ALL found then
{
use ALL
exit
}
}
}
Meaning that from top of file look for noncommented lines and search
for match, DEFAULT and ALL.
As soon as match is found then use it and continue looking *only* for
ALL.
If no match found use DEAFULT if found.
Use ALL in all cases.
Why is this difficult?
And it adheres to the original CVS specification as defined in the
Cederquist. I bet that there are many existing installations that use
this method to fine tune the actions on commit. Why force them to redo
all of their scripting efforts?
/Bo
(Bo Berglund, developer in Sweden)