Re: [Pan-users] ATTN: Testing is gone, all merged into master

[Duncan]

Petr Kovar posted on Wed, 04 Jan 2012 04:03:33 +0100 as excerpted:

> I also fixed some capitalization issues in the master branch.

I just did a git-pull and whatchanged and was looking at some of them... 
and got a new appreciation for your documentation skills! Looks good! =:^)

One thing I noticed in the README... a carryover from the Charles Kerr 
days that needs deleted and/or the feature re-added:

In the SALESPITCH section:

* Pan saves your posted messages in a folder for future reference.

That was old-pan (pre-0.90 C-based).

The way it worked back then, pan had a couple of internal "pseudogroups" 
pan.posted and pan.saved (IDR the specific names of the groups, tho, 
maybe it was message.posted and message.saved, or...).  Since pan of that 
era expired messages when they expired from the server, and these 
pseudogroups didn't actually appear on a server, they didn't expire, so 
both groups were more or less permanent.

As you no doubt know, pan stores messages using the message-id, and did 
so back then as well.  So pan would assign a pre-posting message-id and 
save it to the posted pseudogroup before posting.

Unfortunately, while this was a very clever scheme in theory there were a 
few problems with it.  The biggest was that since pan tracks messages 
across groups and servers via message-id, when the post showed up on the 
groups it was posted to, a user would see it, but it'd already show up as 
"cached" and wouldn't be redownloaded (unless the user manually deleted 
the one pan had saved pre-posting) -- pan would simply display the 
version it had saved off before posting.  As such, it was all but 
impossible to see what was actually posted, for example to check what 
headers the server might have added or to see that it posted correctly.  
If it got mangled, you wouldn't know unless someone quoted the mangled 
version or otherwise mentioned it in a reply.  There were a few other 
corner-case issues with it as well tho I don't remember what they were.

That's why Charles chose not to reimplement that feature, instead 
implementing the drafts feature, allowing people to save off a copy 
(choosing their own name) if they wanted, but not doing so by default.  
But apparently he never removed that bit from the readme, and nobody that 
was familiar enough to catch that it was missing ever caught it until now.


So for now, that bullet point should be removed from the readme.

And if the feature is ever re-added, care should be taken to do it in a 
way that doesn't prevent the poster from downloading what actually got 
posted as well, so the poster can see what actually appears to other 
readers downloading his message.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman