Don't wait for sa-learn

Don't wait for sa-learn

[Magnus Henoch]

[-- Attachment #1: Type: text/plain, Size: 227 bytes --]

I use spamassassin for exit processing of spam and ham messages in my
mail groups.  This takes a bit of time, during which I can't use Emacs
for anything else.  Here is a patch that makes Gnus not wait for
sa-learn to finish:


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Type: text/x-patch, Size: 286 bytes --]

--- orig/lisp/spam.el
+++ mod/lisp/spam.el
@@ -2679,7 +2679,7 @@
 	  (apply 'call-process-region
 		 (point-min) (point-max)
 		 spam-sa-learn-path
-		 nil nil nil "--mbox"
+		 nil 0 nil "--mbox"
 		 (if spam-sa-learn-rebuild
 		     (list action)
 		   `("--no-rebuild" ,action)))))))

[-- Attachment #3: Type: text/plain, Size: 8 bytes --]


Magnus

Re: Don't wait for sa-learn

[Ted Zlatanov]

On 18 Feb 2006, mange@freemail.hu wrote:

> I use spamassassin for exit processing of spam and ham messages in my
> mail groups.  This takes a bit of time, during which I can't use Emacs
> for anything else.  Here is a patch that makes Gnus not wait for
> sa-learn to finish:
> --- orig/lisp/spam.el
> +++ mod/lisp/spam.el
>>> -2679,7 +2679,7 @@
> 	  (apply 'call-process-region
> 		 (point-min) (point-max)
> 		 spam-sa-learn-path
> -		 nil nil nil "--mbox"
> +		 nil 0 nil "--mbox"
> 		 (if spam-sa-learn-rebuild
> 		     (list action)
> 		   `("--no-rebuild" ,action)))))))

I don't know if this is a good idea.  If you submit multiple jobs at
once, does sa-learn do the right thing?  I use a version heavily
customized by my ISP, so I don't know if the vanilla SpamAssassin will
work.

Ted

Re: Don't wait for sa-learn

[Magnus Henoch]

"Ted Zlatanov" <tzz@lifelogs.com> writes:

> I don't know if this is a good idea.  If you submit multiple jobs at
> once, does sa-learn do the right thing?  I use a version heavily
> customized by my ISP, so I don't know if the vanilla SpamAssassin will
> work.

Some source diving tells me that it tries to acquire a lock on the
file, and fails after ten retries (one second between each), so it
should be safe.

Magnus

Re: Don't wait for sa-learn

[Ted Zlatanov]

On 20 Mar 2006, mange@freemail.hu wrote:

> "Ted Zlatanov" <tzz@lifelogs.com> writes:
>
>> I don't know if this is a good idea.  If you submit multiple jobs at
>> once, does sa-learn do the right thing?  I use a version heavily
>> customized by my ISP, so I don't know if the vanilla SpamAssassin will
>> work.
>
> Some source diving tells me that it tries to acquire a lock on the
> file, and fails after ten retries (one second between each), so it
> should be safe.

That seems like a bad situation to me, because the sa-learn run has
the potential to fail silently if the first run is longer than 10
seconds.  Maybe we need to set up a job queue.  That may be hard to
implement, although it will be generally useful.

I'm OK with adding a variable instead, spam-train-in-background, which
is nil by default but can be turned on for specific backends.

Any opinions?

Ted