Kostya, thank you for the detailed explanation.
I haven't had a chance to evaluate this new behavior as I only had 1.8.x and earlier versions installed on my devices, and based on your explanation I hope this design decision will proof to be advantageous for majority of cases. (There've been a few cases where I originally disliked your decision, but liked it when I saw its implementation.)
But I just wanted to comment on one well known paradox related to the seemingly obvious argument such as "
we regularly had complaints about ... " when it applies to a long-existed feature. It is obvious that good developers constantly listen to the complaints (and other feedback) and try to address them. And they should.
However, several well known cognitive biases are hidden here, of which the most important are probably
negativity bias, and
selection/sampling bias: you base your decision on the voice of only those who are unhappy and thus louder.
After you've made the change, you would be getting complaints from those who were happy with the way it had worked before. Based on those, can you correctly evaluate that the number of users who are unhappy with the change? My suspicion that the answer is likely "no!". And there are several reasons for that. There is a potential problem with the "silent majority" in that case, especially if you announce, that this change is ultimate and will never be reverted: Some people don't want to waste their time and energy if they see it's a final decision, and either suck it up or leave silently. But that doesn't necessarily mean that the change satisfied more people than got people unhappy.
(And there is a possible
"change bias" after the change is made.)
And then there is something I'm somehow failing to understand: in order to zoom in, one had to touch the screen, with two fingers, so most likely you're using two hands (holding the phone in one, and using the other hand for the pinch zoom gesture).
What's so hard then about scrolling if necessary?
And then there is vertical scrolling, nobody seems to mind that?
If that was not a rhetorical question but a genuine one, let me try to provide my point of view which is in part based on what I know from cognitive research related to visual design and reading.
First a few facts, rules (not absolute, but defining the baseline), ...; most of these seem to be obvious once stated, but listing them here would help to provide my view.
1. For very short sentences (as in on-screen, say, PPT presentation), one should avoid anything longer than 2 lines (especially in rapid technical presentations). This means that besides avoiding long sentences, it is undesirable to have narrow text-boxes that force extra line wraps.
2. The length of a text line should not exceed a certain viewing angle so that the reader does not need to move the eyes (or even the head) sideways excessively.
3. Every jump of eyes from the end of the previous line to the beginning of the next one is a small disruption to reading. Frequent jumps due to extremely short lines can complicate/slow down comprehension (parsing) of complicated content with lengthy sentences. This can be significant when the line width becomes 1-3 words (as sometimes seen in newspapers, text wrapped around an image). (That's why for longer texts, the line length shouldn't be too short to avoid frequent jumps.)
Horizontal scrolling exacerbates the problem of the necessity for the eye focus to jump across the line wrap, as the eyes loose the line you were reading (skipping ahead, or returning to the already read one).
How does this apply to reading e-mails? In my opinion, it depends on the content and size of the text. If it is only 1-2 lines, it might be more beneficial to scroll horizontally. (That way, you also avoid scrolling vertically mid-sentence.)
For a [significantly] longer text, narrow lines (= higher frequency of jumps) become less of an evil compared to horizontal scrolling. (Of course, this is until the line is so narrow that many single words get wrapped. For that reason, when I need to zoom in and the text is reflowed, I am usually trying to switch to the landscape format.)
Also, for long texts, you already need to scroll vertically; that's unavoidable. Scrolling in 2 direction while reading is much more disruptive for tracking the text with your eyes, puts more strain, requires more attention for parsing and comprehension.
PS. An aside Re: vertical scrolling.
The was a recent switch in the website design paradigm, where the earlier concept of avoiding vertical scrolling by having multiple pages has been replaced with long (sometimes "infinite") vertical scrolling. I don't know if it is based on a noteworthy research, or just a fad.
And previously we had reflow and now we don't... and I don't think there are many email apps that can reflow HTML messages.
That was one of the advantages of Aquamail. I just read a message in Gmail app (I have one account still using Gmail app, because Aquamail doesn't have per-account configuration for syncing/not-syncing on mobile network, and I need only one account to sync and notify quickly on
any network.) - it was a long message and I once again confirmed that I strongly dislike scrolling horizontally.
There is actually one additional problem with that: swiping accidentally to the next/previous message (or pulling out the "side cabinet" in case of Gmail -- that's not a problem in Aquamail).
Speaking of reflowing in other e-mail clients, - I don't know about Android based once, but it's a standard feature of Thunderbird on Windows (both when you zoom in and when you just change the window width). And yes, when you zoom in, the focus changes, and an image/text can move aside. But it is seldom a big problem (there or in Aquamail of many years).