I have been having this sort of 'random crash' of Pegasus Mail for Windows 4.whatever for a long time myself - most notably when the systems appears to be rendering some 'graphical' email or another.  After booting the same message will invariably display. It does get very frustrating at times.

 Will this new version of Bearhtml be available soon? 

For the past few months Pegasus has been crashing unexpectedly for no apparent reason. Some times it crashes when you click a url link, other times when I click on a message to preview it (same message preview after I restart shows the message fine, so it can't be the message itself), othertimes when I click to save a file or move a message to a folder. All seem to happen when a message is in the New Mail folder.

I have 200+ messages in the New Mail folder. Could that be the cause? Or is there something else?

The random nature of it makes it difficult to replicate or troubleshoot, I am afraid. In the past week or so I have encountered a crash more than once a day. Sometimes rebooting the PC alleviates the problem.

 Any clues? I am using a Dual-Core PC, albeit I doubt that would have anything to do with it.

[quote user="Valerio"]

 Any clues? I am using a Dual-Core PC, albeit I doubt that would have anything to do with it.

[/quote]

It could if WinPMail is not bound to just one of the processors.  Use the  Windows Task Manager and right click on the WINPM-32.EXE process.  Use "Set affinity" to force WinPMail to use one of the processors.

Thanks I will try that.

I have the winpm32.exe process set to use both cores in my dual-core AMD system.  I don't see any evidence of crashing, so I doubt that this is the problem.  Unless Pegasus is multi-threaded and lacks process interlocks (or barriers) then why should dual CPU's be an issue?

[quote user="jss1941"]I have the winpm32.exe process set to use both cores in my dual-core AMD system.  I don't see any evidence of crashing, so I doubt that this is the problem.  Unless Pegasus is multi-threaded and lacks process interlocks (or barriers) then why should dual CPU's be an issue?
[/quote]

Believe me it is an issue with some CPUs and systems.  WinPMail is NOT designed to use multiple processors.  It will work but if you are having problems with crashing then the use of a single CPU many times will solve the problem.

'

Actually, WinPMail should have no problems on multi-CPU systems, because it's not heavily threaded at all (it will be in future, but will have proper interlocks).

There *IS* a long-standing problem in MERCURY with multi-CPU systems which I've never been able to track down... The "solution" in the end was to force the program internally to use only one CPU, which isn't perfect, but is a good enough solution for the most part.

Like I say, though, Multiple CPUs shouldn't be an issue in WinPMail. What you're describing sounds more like a problem Martin Ireland and I have been dealing with involving certain types of malformed URL in HTML mail. I believe Martin has a version of BearHTML (the HTML renderer Pegasus Mail uses internally) that should deal with this problem, but I'm not sure when he might be making it available.

Cheers!

-- David --

Yes, I have a new version of Bearhtml  (3.2.0) going through initial testing right now. It is a major rewrite of the Html validation process. I have put it through my gallery of problem messages, and I have not had one crash so far. By the way, that doesn't mean that a message gets displayed. Some Html is *so* bad that the agent cleaning up Html, Tidy, aborts completely. I am now trapping this event to stop it bringing down the whole of Pegasus Mail.

My next step will be to offer it out to the beta testers for a couple of weeks, then issue it out to the whole community, via this community website.

But as you know, Murphy is alive and well, and I never cease to be surprised at the broken attempts at simple html streams. So time will tell

Martin 

[quote user="irelam"]Some Html is *so* bad that the agent cleaning up Html, Tidy, aborts completely.[/quote]

Good grief ... although, I guess, not surprising:  Tidy when used at the W3C website (or however else), will sometimes put up a message that it's been offered mark-up with, for example, so many proprietary extensions it can do nothing with it.

One supposes that the relative leniency of HTML parsers in the past was probably a factor in the rapid expansion of the web: anyone could put a page up, and, within limits, it could be read even if it were pretty horribly written.  But, goodness, it's led to bad mark-up being almost more the rule than the exception.  It's said that many of the XHTML webpages out there would break if actually served as application/xhtml+xml (as at least some versions of XHTML really should be) rather than as text/html, because XML parsers aren't, and aren't supposed to be, "forgiving".

This may or may not be relevant to this particular problem, but is a pattern of Pegasus crashes that I have lived with for a long time.

It started with Win 98, and the current PMail versions of the time. The only failure ever with Pegasus was the familiar 'invalid page fault error', and it almost always came three in a row, one immediately after another.

For me, this signalled that Windows was gradually getting its knickers in a twist, and that it was time to reboot my 24/7 working machine.

With XP and Pegasus v4.41, the same pattern continues, except of course that Bill does not tell you what the problem is - all you get is "the program has encountered a problem and must close...". However, I know that this is the same old invalid page fault error. How? Pegasus is the only application that has ever produced three in a row, and continues to do so under XP.

As a programmer, I understand page faults and their place in the Windows scheme of things, so it has never been a serious concern.

Aris

IMHO the problem is that there are *very* few decent Html editors out there. Of these there are a small percentage that will do a good job of Cut-and-paste. As an example, one message I received from a Pegasus Mail user had 5 "Html" tags in the "Body" portion of the html stream. This contravention of protocol can be ignored gracefully, but it indicates lack of working knowledge of the current editor environment.
In this particular case, this causes all the existing Styles and CSS definitions to be cleared which makes a real mess out of html that is trying to achieve a complex display.

Martin