Genii Weblog

Civility in critiquing the ideas of others is no vice. Rudeness in defending your own ideas is no virtue.

Mon 2 Feb 2009, 11:47 PM
The rules are simple.  Without testing before hand, make a prediction about a MIME rendering problem (in other words, mail getting sent or iNotes mail being read that isn't sent or read well), preferably one I haven't already identified.  For example, I'll make a couple more and then test them (although perhaps not until tomorrow, as it is getting late).  I should point out that two huge weaknesses are part of most of these predictions.  The first is that when a rich text message is shown through iNotes, it does not use the MIME rendering engine, but instead uses the web rendering engine.  The web rendering engine assumes that the Domino server is available, but an email going to another recipient can't assume that the original Domino server is available.  The second weakness is that the various web and MIME rendering engines use virtually no CSS, inline or otherwise, and rely entirely on HTML.  Use those clues in making your predictions.

1) I predict that if someone sets a section title to white lettering on a dark background with one of the various filled in section types, such as one of those shown below, the title will be left white but the dark background will disappear, making it look like there is no title to the recipient.  I predict this will be a problem with both iNotes reading and server MIME rendering (because we already know from before that client MIME rendering loses the title completely).

White lettering on darksection backgrounds

2) I predict that when a  file attachment is included in a rich text message, and that message is viewed through iNotes and then forwarded, the attachment will be missing from the forwarded message, but the broken link will remain.

OK, now it is your turn.  Just close your eyes and imagine some rich text and how it could be rendered wrong if the developer did not think hard about the implications and possibilities.  It is OK in this case to play "gotcha!" by imagining things like the white letters on a dark background that I mentioned above.  It is NOT OK to use predict problems with buttons or scripts or that sort of thing, as most active content is stripped out on purpose.

I will do testing with the very latest production version, either Notes or Domino 8.5.0, and let you know the results of my predictions as well as yours.  C'mon, show me how creative you are!

Copyright 2009 Genii Software Ltd.


Mon 2 Feb 2009, 10:51 PM
I don't know whether this is exciting or depressing.  Look back at the prediction I made a little bit ago.  I did a bunch of careful tests to see what happens, and what happens is just about what I predicted, at least for tabbed tables.  Sections are handled better, I must admit, which does make sense given their frequent use in emails for reply sections, so the challenge was only half met, I guess.  Anyway, here are the screen prints to show the problem with the tabbed tables.  After forwarding, only the open tab will be accessible to the recipient, and the other tabs are just broken links.

Tabbed table in Notes client (same as I used before)

Tabbed table in Notes client

Tabbed table in iNotes web client (see the predicted hotspots)

Tabbed table in iNotes web client

Tabbed table in iNotes after forwarding (links broken, since they require round trip to a server that isn't available to recipient)

Tabbed table after forwarding in iNotes

Copyright 2009 Genii Software Ltd.


Mon 2 Feb 2009, 09:49 PM
Sometimes I feel like I MUST be missing something.  Mail has been a critical part of Lotus Notes for two decades, so obviously there must be someone, or perhaps multiple people, at IBM whose job it is to think... think about potential issues... think about scenarios where mail might not get through successfully... think about the customers who have to be able to use the system with confidence.  But sometimes it is hard to believe.  More often, I simply think "What would happen if somebody wrote this who didn't bother testing - what would they miss?" and then I go test it.  Usually I find a problem, although sometimes not exactly the one I expected.

My challenge: Predicting an untested problem
I'm going to try an experiment.  I'm going to predict a problem before even trying it out, then test to see if I am right after I post this challenge.  I have no way to prove to you that I have not already tested this, so you will have to trust me.  To hedge my bets a bit, I will make it a bit of a broad prediction.

My prediction: iNotes will render tabs and sections as if on the web
I predict that if I have a tabbed table or section in a rich text email, and I go to iNotes to view the email with the tabbed table or section, it will render as if it were going to the web, with active hotspots to switch tabs or open/close the section, rather than as if it were sent by mail, with a flattened table or opened section.  Then, when I forward or reply the message from iNotes, the hotspots will remain but have lost their meaning, so that the tabbed table will only show one tab, or the section will remain closed if it was closed in the rich text message.

Watch for my follow up to see if my prediction is correct.

Copyright 2009 Genii Software Ltd.


Mon 2 Feb 2009, 10:37 AM
It would seem to me that these two ideas would have similar appeal, but the response to one has been much more robust that the response to the other.  I am curious whether that is because:

a) email fidelity is more important to people than web fidelity; or
b) web designers have pretty much abandoned using native rendering and control everything with pass-thru HTML; or
c) Gregg and I have both given lots more publicity to Gregg's idea than Lisa has to hers; or
d) none of the above.

So, while I'd encourage folks to vote for both of these ideas, I'd be very interested in whether Lisa's idea also is of interest to people (a potential AppFidelity product, perhaps?)  Anybody have any opinion on what the reason(s) might be?

Copyright 2009 Genii Software Ltd.


Mon 2 Feb 2009, 08:43 AM
Part of my reason for having this blog is to share the real life experiences of running a small software company.  While there are wonderful advantages to being nimble and not having committees decide everything, there is the disadvantage at times of being your own QA department.  While Quality Assurance is critical, and something I take very seriously in shipping products, I sometimes slip up and forget to properly focus on Quality Assurance in beta versions.

So, on Friday, after a great deal of testing of the first beta of our  iFidelity for Lotus Notes  product, I should have known not to make that one last tweak.  The tweak made stress testing easier, and seemed like a perfectly safe and sensible, though minor, performance enhancement, but due to misplaced parentheses (I know how some sad person at Google feels right now), messages were getting converted to MIME that shouldn't have been, then getting converted back to rich text using the native MIME to CD conversion.  The result was ugly messages, the exact opposite of what was intended.  While it was a reminder that if the Notes inbound conversion were not so bad, this would not have been such a problem, that is cold comfort.  I still broke a cardinal rule to not make changes right before sending out a version, beta or not.

So, mea culpa, mea culpa.  Just don't tell Paul.

Copyright 2009 Genii Software Ltd.