Flyspray - The bug killer!

I’m not an expert on this field but I have had similar thoughts in other projects. One thing to consider on most server configurations is that there is no way to execute a script on mail reception, this means that someone has to manually access a script that checks for new mails and inserts the task.

Another question is; what’s the benefit? If you have mail access, you most certainly have access to flyspray online. Furthermore, you must specify the different fields like ‘Task type’ etc in some way.

But I agree, it would be a neat feature

Would you autogenerate the mails from another system?

You might take a look at the Wordpress code, which has such a feature (I’m sure others have too)

this can be done with the XMLRPC interface (no need to look WP code, it is awful :-P)
However, Im not going to implement this.. looks easy to do, except one part “and if it comes from a recognised user..” mail headers are sooooooooo easy to spoof

Florian what you think ?

I think that we should implement this feature because it can be really handy. Especially nice would be if one could reply to comment notification emails to add a comment.

Authentification is a problem of course. The email address alone is not enough, so either the username/password combo has to be included to the email, or each other gets an autogenerated key for use in emails so that the login data is not transferred in plain text.

let’s see

Alternative one.(just dreaming )

1. a comment (or a inital bug report), generates an email, this mail contains an unique, non constant, exclusive random hash , valid only for one reply, which is associated to the address which the email was sent(that is one hash per notification).
2. a command line script attached to the correspondent mta facilities parses the message and validates is authorized,and then talks flyspray via xmlrpc. hence creating the comment.

alternative 2:

- A plain clear bridge between mailman and flyspray.(I vote for this :D)

I would just write next to the user name in a bug report or a comment “sent via email”.

This shows that it’s not totally secure who wrote the comment/bug report, but if it is a problem, it can be deleted/commented/discussed anyway.

This would be the easiest solution IMO.

a comment (or a inital bug report), generates an email, this mail contains an unique, non constant, exclusive random hash , valid only for one reply, which is associated to the address which the email was sent(that is one hash per notification).

But this wouldn’t work for other email actions.

a command line script attached to the correspondent mta facilities parses the message and validates is authorized,and then talks flyspray via xmlrpc. hence creating the comment.

Fist problem: No one maintains the XML RPC code atm. Second problem: Command line script doesn’t sound good. If we implement this feature, it should be possible for everyone to easily use it without access to the command line.

This shows that it’s not totally secure who wrote the comment/bug report,

I will never implement “not totally secure” features on purpose.

Florian.. and how the hell you pretend to implement it without command line script... ? o_O

email message has to be processed by the mta and then something should parse the email, remove all uneeded parts and add to flsypray only message body...

I suggest that the admin sets up a mailbox (any provider) where all emails @Flyspray can be sent to. Then Flyspray periodically checks (can be added to the reminder stuff) the mailbox for new emails and processes all emails that contain commands for Flyspray. This way everybody can make use of it without additional software or permissions on the server.

I am currently writing an extension for a forum software I use, that performs many of the things that are mentioned here. When I get that done, I will let you have a copy of the code to see if some or all of it could be applied here.

the people over at mantis (which i have to use for work because of our webhost) are also working on something like this. the plan they came up with is very similar to how flo is describing.

if i get a chance i can look into churning something out.

Mantis Bug Report

Hi, I registered on the BTS just to vote on this feature request!
I would LOVE email bug reporting.

Most end users tend to find bug reporting systems rather scary so I would love to have people email their problems to bugs@mydomain.com and it would show up as a task.

Even if there are no commands in the email (such as category, severity, etc) the bug report would show up in flyspray and one of the admins could then complete the information.

Zentrack [http://www.zentrack.net/] has it!
Egate Overview: [http://www.zentrack.net/modules/cjaycontent/index.php?id=5]
Scheduled task (cron) can check an email account.
Setting up email interface: [http://www.zentrack.net/modules/cjaycontent/index.php?id=6]
You'll need Mail_Mime for this.
HTH

Ok, the only correct/viable/not too complex way to do this is like the comment 2799 by floele. that will work. ;)

Bugzilla put this into version 3:
http://www.bugzilla.org/releases/3.0/new-features.html#v30_feat_emi

From http://www.bugzilla.org/docs/3.0/html/api/email_in.html :
CAUTION The script does not do any validation that the user is who they say they are. That is, it accepts any ‘From’ address, as long as it’s a valid Bugzilla account. So make sure that your MTA validates that the message is actually coming from who it says it’s coming from, and only allow access to the inbound email system from people you trust.

Doesn’t look like a solution for us. Most people certainly can’t configure their MTA (both because of lacking knowledge and permissions).

The outcome so far:

http://groups.google.com/group/flyspray/browse_thread/thread/a931b02426a5564e?hl=en

So in other words, we’ll most likely drop this feature.

I find the reason for closing a bit close-minded.
I pipe email through a gmail or google apps email account, and it's 95-99% free of spam.
Also the issue of spam is separate from the need for email pipelining.
All enterprise and well-developed support systems offer email pipelining - it's probably the #1 feature of all support systems.

Think about your logic.
There are annoying potholes on roads, therefore we don't drive cars.