[imp] Big attachment problem

[Michael M Slusarz]

Quoting "Kevin M. Myer" <kevin_myer at iu13.org>:

| See http://bugs.horde.org/show_bug.cgi?id=1162 - lots of finger pointing,
| IMHO
| about where the real problem lies, but it should be enough to understand
| whats
| going on.  The problem is NOT that the machine isn't powerful enough to
| handle
| large attachments - its that as IMP uses PEAR's SMTP routines, the size
| of the
| attachment balloons in memory to a huge number (many times more the
| original
| attachment size), which quickly hits any sane PHP memory_limit size.

No, the solution is that you *limit* attachments to a certain size.

First, this is not the problem that is being asked about in this thread - 
here, the user is not being able to send any attachments, not that he can't 
send big attachments.

Second, as mentioned in the bug report, the above behavior is simply *NOT* 
a bug.  It may be inefficient, but it is not broken.  This is like saying 
AutoCAD is buggy since it will not run on a machine with only 64 MB.  
Simply put, you _do_ have to use a more powerful machine if you want to 
send large attachments - it's as simple as that.  If you do not want to 
use/purchase new hardware, then a solution is to simply limit attachment 
size via php.ini to something smaller (e.g. 5 MB).

I renew my objection in the bug report that people really, really, really 
shouldn't be sending 19 MB attachments through the mail system.  This is 
not what the mail system is designed for - there are other protocols out 
there that are designed specifically for the purpose of transferring files.

If someone wants to rewrite the PEAR SMTP module to be more resource 
friendly, I'm sure the maintainers of the module (which chuck and jon 
appear to be 2 of them) would be more than happy to commit it.  There is 
definitely no need to write a Horde specific SMTP module, as was suggested 
in the report.  This a 'performance' request, not a 'bug' request, so, 
since I 1) am not getting paid, 2) don't have time, and 3) it works for me, 
don't look to me to do anything about this (soon).  Therein lies the joy of 
open source development... :)

michael

______________________________________________
Michael Slusarz [slusarz at bigworm.colorado.edu]
The University of Colorado at Boulder