What happened to hiring an architect?

Ben Bucksch ben.bucksch at beonex.com
Wed Dec 21 07:14:04 UTC 2016



Am 20. Dezember 2016 20:18:03 MEZ, schrieb Disaster Master <disasterlistmanager at gmail.com>:
>On 12/20/2016 1:48 PM, Ben Bucksch <ben.bucksch at beonex.com> wrote:
>> I don't want Thunderbird to make the same fatal mistake. Slipping
>into
>> it is one thing. Running into it consciously is just inexcusable.
>
>
>> I'm saying: we need to accept that fact that foregoing gecko is not
>an
>> option. Unmarrying thunderbird doesn't work either.
>
>You left out the other option - rewrite the parts of Thunderbird that
>rely on the deprecated Gecko components.

That is the "unmarrying" i mentioned. 

The whole front-end is written in XUL. So, you're rewriting the frontend.

The backend is based on XPCOM. So,  you need to at least refractor that. 

Ther isn't much left to salvage.

>
>I honestly want to know why you seem to think that would be impossible
>-
>or even harder than writing an email client from scratch that has
>reasonable UI and feature parity with current Thunderbird.

Because i did both.

I've worked on thunderbird since 1999, on backed - Plaintext handling in libmime; authentication logic in the IMAP, POP3 and SMTP protocol implementations - and  UI - the new account wizard.

For another client, in another project, I also reimplemented parts of IMAP and POP3 and a full XMPP stack in JS. I implemented that mostly alone, so I have an idea of how long such an endeavor takes. (BTW: I'd be happy to contribute the XMPP implementation to such a new project,  as i own the full rights to it.)

So,  i know both from first hand experience. 

>
>> So, that leaves only one choice: Write a new client. Yes, that will
>> take 1-3 years.
>
>Really. You believe we can have a reasonably close facsimile of current
>Thunderbird, with the resources we currently have, in 3 to 5 years.
>
>Kent said it would be more like 30 man-years.

The 2 figures are not to far off. I'm talking about calendar time,  kent about work time.  I assume 10 devs on it,  so that's almost the same figure. I do think that's realistic and feasible. 

I also count in that a very experienced and efficient dev can be 5 times more productive than entry level devs. That's my experience from leading a big commercial XULRunner application.

I also think that thunderbird has or can gather the necessary resources.

-- 
Sent from my mobile phone. Please excuse the brevity.


More information about the tb-planning mailing list