Windows Mobile Web Browsing Bible - PART II

CONTINUED FROM HERE – CLICK THE LINK if you haven’t come from there!

4. Comparison / feature chart

It's available HERE. It also contains some 360 screenshots, almost all taken on a WM6 VGA HTC Universal (don’t forget to click the links to see them if interested)!

As with all my feature charts (and roundups), I’ve paid special attention to provide you with mini-tutorials when discussing a particular question. For example, when I elaborate on the “One column” mode (see the “One (single) column view?” row in the chart), with, say, Minimo, I also show how you can actually switch to this mode by showing a screenshot of the menu item taking you there. This means the chart contains hundreds of small, but, in cases, very useful quick tips & mini-tutorials you won’t find anywhere else. All in a very compact form: just imagine how much I would have ended up having to type upon trying to convey the SAME deal of information in a non-tabular form – yeah, dozens if not hundreds of kilobytes.

Of course, I have tried to be as verbose and clear as possible when explaining the different test cases. I’ve also paid special effort to linking in my previous, related articles on the different tests I’ve conducted. For example, when I provide a link along with the Internationalization support group, it means you may want to follow the link to find out what the tests in this group are all about.

4.1 Explanation for the Comparison / feature chart

(Note that all browsers support SSL (secure connections); therefore, I haven’t included this in the chart, as opposed to the previous version of this Bible (at that time, Minimo still didn't support SSL). Note that Opera Mini has only recently, with the 3.x series, received support for SSL.)

Platform compliance? group: in here, I've elaborated on the operating system compliance of each and every browser. I've grouped together the platforms that, compliance-wise, behave the same way. That is, a WM2003-compatible program will surely run on WM2003SE; a WM5-compatible program on WM6. I've also noted the exceptions or some problems; for example, with SPP. Also noted is the lack of support for newly introduced PIE features like One column in WM2003SE, Save images / Full Screen in WM5 and Use High Resolution in WM6 VGA.

It's no news older platforms are all phased out - and this, unfortunately, already means completely losing support for relatively new operating system versions like WM2003SE. NetFront 3.4, Minimo 0.2 and DeepFish are all WM5+-only; so will be the forthcoming Opera 9. However, older versions of these browsers (except for, of course, DeepFish) do/did support WM2003(SE); in the chart, I've mentioned the actual version number that still did this. Support for the now-ancient Pocket PC 2002 operating system is even more scarce; of the new releases, only PIEPlus and SPP support it. Finally, non-ARM-based Pocket PC (2000) devices are completely abandoned.

Screen group: in here, I've elaborated on the different screen resolutions (QVGA, VGA, square) and orientations (Portrait and the two Landscape modes). Fortunately (except for the complete lack of support for square screens in Thunderhawk), current Pocket PC browsers are all VGA (including native (non-SE) VGA modes) and Landscape-compliant, where the latter also includes left-hand landscape modes used on WM models with built-in slide-out / clamshell keyboards.

Screen estate utilization group: everything related to how browsers are able to make use of the available screen estate.

Full screen mode?: can you switch to full-screen mode, hiding the taskbar at the top and the command bar at the bottom? I’ve also noted the way to switch back to normal mode; it’s, for example, a little icon as with all the three (real) PIE plug-ins, which is the best and least space-consuming.

As can clearly be seen, Opera Mobile, Minimo and NetFront all display the tab bar (and, with Opera Mobile, the address/icon bar) even in full screen mode. This is certainly a drawback.

Address bar hiding?: in pre-WM5 PIE's (as with several other browsers), you could hide the address bar to free up some screen estate. In here, I've scrutinized whether you can do the same in the reviewed browsers. Note that Opera Mobile displays the combined address bar / command bar even in Full Screen mode, which should be addressed in a later version.

Scrollbar (may be) hidden in full screen mode?: better browsers and browser plug-ins may be configured to hide the horizontal/vertical scrollbars in full screen mode. Unfortunately, only MultiIE and PIEPlus support this; Opera Mobile, Minimo, NetFront and PIE (without either PIEPlus or MultiIE) don't.

Context menus group: while I've also dedicated separate rows to elaborating on mostly context menu-based functionality like opening a link in a new tab (instead of the current one), saving an image or copying a link target address to the clipboard, I've also chosen to collect screenshots and a quick list of the additional, new context menu items available with all the three different entities in a Web page (not counting in special entities like Flash animations, Java applets or frames; with the first two, there are no context menus; the latter is scrutinized in the Frames group): images, links and generally non-image/non-link content.

Advanced address bar features (macros, completion) group: this section lists the different types of macros and address bar (auto)completion. The rows and screenshots in this section are pretty self-explanatory; therefore, I don't explain them in here.

Rendering modes group: the screen resolution of a Windows Mobile device is inherently smaller than those of desktop / notebook computers. Even the largest WM screens (800*480 in, for example, the new Toshiba G900) are still smaller than the XGA (1024*768) screens used in even basic notebook models, let alone higher-resolution ones (for example, I'm writing this article on my UXGA (1600*1200) Thinkpad.) Low-resolution WM devices with either a QVGA or a square screen are even worse.

With these low-resolution screens, it's pretty understandable a Web page can't be correctly rendered in its original layout. A layout designed for a horizontal resolution of at least 800 pixels just can't be correctly rendered on a screen with a width of 240 pixels. This results in (mainly) three approaches:

1. render the page as is, in its original layout - that is, make the user scroll horizontally. This is the worst approach as you will end up having to scroll horizontally to read each and every row.
2. while trying to keep the original horizontal layout, try to resize every horizontal page entity so that they fit in the screen horizontally. This approach, in general, works OK on VGA devices, particularly when used in Landscape orientation (that is, with 640 active pixels, even when you subtract the width of the vertical scrollbar). On the other hand, with QVGA screens (and particularly with square ones with a meager 240 pixels), this approach wont really work because, in some cases, each column will only have space for 3-4 letters at most. (See the examples in the first row of the group showing this in practice; or, the NetFront Just-fit example showing a QVGA screenshot in the earlier version of the Web Browsing Bible!)
3. finally, try to render all cells in a row of a table or all frames vertically; that is, one cell or one frame a row.

Note that there may be combinations of the latter two approaches; NetFront's Smart-fit rendering is a perfect example of this (using the most recommended Full Rendering mode). It, when it notices that there simply are too many for example table cells in a row, makes sure it renders all of them vertically. When, however, it notices somewhere else on the same page there isn’t enough screen estate, it will render the cells in separate rows. The PPCMag test example, used throughout the entire chart for testing, is a perfect example of this. At the top of the page, where there are only two text input fields and some text, these are shown in the same row (unlike with "real" One column solutions). However, with much more information / text in a row (the case with the body chart itself), most of the cells are aligned vertically. This approach unifies the good sides of both approaches and should be implemented by at least the Opera Mobile folks as, say, a fourth way of content display.

The first two rows in this group compare the applications' ability to fit the contents of a Web page (horizontally) on the screen and to render the page in the One column mode, if possible.

Fit-to-screen (tested with the PPCMag test)?:

As can clearly be seen, PIE has always delivered pretty bad results, unlike with all the comparable and fit-to-screen-capable alternatives (except for Minimo, where SSR doesn't always work). Both NetFront's "Just-fit rendering" and Opera Mobile's "Default" mode are far better at really crunching the horizontal contents of a Web page to the available screen estate and, in most of the cases, are perfectly usable on especially VGA devices.

Minimo's SSR mode (whish is enabled by default) is a different animal - it doesn't work with many sites (see the RedHotPawn example). When it does work, however, it also delivers good results.

Opera Mini doesn't have a comparable rendering mode at all (as it's solely using an One column mode). Finally, Thunderhawk renders the page using the original layout, which is pretty much OK in most cases.

One (single) column view?:

As can be seen, the reviewed apps use vastly different approaches. The best approach is, without doubt, that of NetFront for the reasons outlined above. It's closely followed by all native One column-capable browsers: PIE in WM2003SE+, Opera Mobile (particularly now that, with the brand new, 8.65 version, the old bug with the limited horizontal column width has been fixed) and Opera Mini (incidentally, the latter doesn't have other rendering modes at all).

As has been pointed out, it's only with WM2003SE and later WM operating system versions that the built-in PIE supports the One Column mode. In earlier operating system versions, should you really want to have One column rendering and still want to stick with PIE (while, of course, Opera Mobile is far better a choice on WM2003), you will want to take a closer look at PIEPlus, the only PIE plug-in to force the incoming Web content into one column.

Note that you can achieve the same effect with ANY browser using external one-columnizer services like Skweezer, MobileLeap and the like. However, they may result in some problems (for example, because they also get rid of JavaScript code); therefore, you may still want to go for something else.

Rendering mode (does it show the start of the document even when it’s not entirely downloaded?): this test elaborates on how the given browser loads a new document: does it start rendering it only some 2-3 seconds before fully finishing the download (that is, does the user face an empty screen for, say, 90% of the download), or, does it try to render the page as soon as possible?

As can clearly be seen, there are two types of browsers: one set of them (PIE, Opera Mobile) will start rendering the page as soon as possible, while some wait until the download & parsing is almost entirely done (Minimo and the proxy server-based solutions). NetFront is a strange animal because in the normal Full Rendering mode it sometimes delivered very good (starting to render right at the beginning), while, at other times, pretty bad (starting to render only later) results.

Note that NetFront also offers a "Rapid-Render" mode, which guarantees the content will be rendered during page fetching. I can't at all recommend this mode, however, because of the HUGE time overhead, which is particularly an issue in the new, 3.4 version, where the difference in time needed for page fetching can easily be fivefold. Furthermore, the rendered contents you're presented aren't the final ones; they will only be presented later, after a really distracting full screen clear. This may be pretty annoying for the user because he or she may even forget where he/she was and/or will have to scroll around a lot to find it.

Multiple page operation (multitabs) group: in this group, I've elaborated on how the application handles multiple pages; is it, for example, possible to open a link in a background page for background download, and, then, get notified when it's downloaded. All this in order to avoid having to waste time on waiting for the page to be downloaded, which is especially important with slow connections.

Feedback on page loading events (sound effects / bringing to the foreground)?: A decent browser should notify the user when a page has completely been downloaded and rendered in the background. For example, the desktop Opera browser turns the color of the text on the tab where download has ended to blue, which is very easy to notice, even with disabled sounds. In here, I've listed how the tested browsers behave in this respect. Unfortunately, the Windows Mobile version of Opera doesn't do the same trick as the desktop one (and not any sound notification either). This is the case with all the other "real" (non-plug-in) browsers too. Actually, it's only PIEPlus and MultiIE that lets for configuring what should happen in these cases. Kudos to their developers!

Opening links in…-support, particularly as opening something in a background tab is concerned: in here, I've listed whether it's possible to open a link in a new and, particularly, in a background new tab in order to avoid having to manually switch back to the current one to continue reading it while the requested page is loading in the background.

As can clearly be seen, some browsers don't let for background link opening at all; unfortunately, Opera Mobile, NetFront and Minimo also belong to this group. Actually, it's only the three "real" PIE plug-ins that offer background link opening capabilities.

Max. number of tabs open at a time?: die-hard Web browser users may want to prefer having as many pages open as possible. Most browsers and PIE plug-ins do let the user do so; the most important exception is NetFront, which only lets for opening up to five tabs. This is far from perfect and you'll run into this restriction pretty easily if you often open a link in a new tab.

Something should also be emphasized. The Windows Mobile operating system, as of now (the WindowsCE 5.2-based WM6; it's only the brand new WindowsCE 6 and the forthcoming WM version based on it that (will) have got rid of this restriction), doesn't let for more concurrent processes than 32. Most of the reviewed applications (except for, for example, Opera Mini), however, create a separate process for each tab. This means, depending on the operating system used and the number of other programs you run, in general, you can't have more than 20-28 tabs opened with a browser before these start to be terminated (which, in cases, may result in terminating all the browser processes at once - that is, ALL your tabs will be suddenly closed). Again, this restriction doesn't apply to Opera Mini - with it, I had 30 pages opened several times without any problems.

Note that, as both opening new tabs (at least with PIE plug-ins; with non-PIE-based browsers, the memory consumption in these cases isn't at all bad) and rendering Web pages (which is an issue with several Web browsers; most importantly, with PIE) may be memory-intensive operations, it's highly possible you fill up your dynamic RAM program memory much faster than reaching the process limit of the operating system. With the least memory-hungry application, Opera Mobile, I've had no problems in browsing some 27-28 tabs at a time under WM5+, however - that is, you can make good use of your dynamic memory very easily.

Tabs constantly on the screen, their taking up screen estate etc. group: while the previous group didn't concentrate on the visual representations of the multiple browser document windows, this one does. In here, I elaborate on whether you can alter the tabs' size (and their taking up valuable screen estate, particularly in the Landscape orientation), whether they're displayed in full screen mode, whether you can configure the system to open the new tabs next to the current one, or, strictly at the end of the tab list; whether the tabs have context menus (in this respect, Minimo is clearly the best) and, finally, whether the tabs can easily be closed with, say, only one screen tap.

Misc. group: the tests in this group speak for themselves. Please make sure you consult the screenshots, should you still not get the point what they are all about. I only elaborate on the Access to standard PIE favorites? group, which shows whether the given browser is able to access the PIE favorites for either reading or writing, or both.

As can clearly be seen, while the traditional file system representation of favorites is very simple to handle, only three browsers have support for it: Thunderhawk, Opera Mobile and NetFront. This means neither Minimo nor Opera Mini have support for them. (The latter is, of course, understandable, taken the restricted “sandbox” midlets are provided, file access-wise.) Furthermore, Opera Mobile isn’t able to create PIE-compliant favorites (not even when you create these favorites explicitly in the Internet Explorer Favorites folder); this means favorites added in Opera Mobile will not be visible to other browsers and you can’t synchronize them back to your desktop computer(s) either.

Note that the WM operating system also stores favorites in the Registry; both NetFront and Opera Mobile were able to read these Registry-based favorites.

Standards compliance groups: in the five groups here, I examine the following four areas (and a miscellaneous area with some "not suitable for bigger groups" tests):

JavaScript, scripting, Java (Part I) : in here, I've run several tests to find out the compatibility of all the browsers with some well-known pages having very strong and complicated scripting. As can clearly be seen, Opera Mobile and Minimo have by far the best JavaScript and AJAX support. PIE has always had a very bad JavaScript support and, even as of WM6, non-existing AJAX support. (Frankly, I don't understand why Microsoft states PIE in WM5 AKU3 / WM6 is AJAX-compliant, when it just isn't. Its JavaScript compliance isn't a tad better than in older versions either. There must have been a misunderstanding somewhere.) NetFront had mediocre JavaScript support in 3.3 and good in 3.4; as far as its AJAX compliance is concerned, 3.4 was indeed a BIG step ahead (albeit it's still worse than that of Minimo or Opera Mobile).

Finally, it's in here that I also elaborate on the Java applet compliance of the Web browsers. Unfortunately, Minimo and the two Operas have absolutely no Java support. This isn't that big a problem, however, because very few sites do actively use Java applets - it's mostly Flash that everything is based on (see Flash compatibility later).

Thunderhawk and NetFront both have their custom Java support, which can't be swapped to something else. With PIE, however, you have some choices when choosing a JVM: CrEme and the no longer sold / supported JEODE, which, back in 2001-2003, was shipped on iPAQ CD's. Of the two, I'd prefer CrEme because of the vastly superior speed and generally better compliance. The reader is kindly referred to my other, related articles for more information on CrEme.

HTML Frames: these test concentrate on the frame support of the Web browsers. You may have already heard of PIE's only supporting few parallel or embedded frames and absolutely not supporting so-called "Iframes". In here, I elaborate on all these issues. If you know a bit about HTML and would you find out how I've did the tests, don't forget to check out the HTML test pages I've created for these tests: I've linked in them all. They're pretty instructive.

As can clearly be seen, Opera Mobile has the best frame support when it comes to the maximal number of parallel / embedded frames. Its only problem is the lack of "go to a frame" functionality (to maximize a given frame to the entire screen), which, otherwise, would be REALLY important, particularly when you really wouldn't need the contents of the other frames. The Opera folks will want to address this issue. PIE, on the other hand, is at the other end of the spectrum: its frame support is the worst of all, frame number-wise.

Finally, some really good news for PIE freaks: in WM6, Iframe support has finally been added. It's not really flawless (see my comments and the screenshot), but, at least, it's already there.

Internationalization support (Part IV): please see this article for a complete description of what this all means.

Finally, the fifth subgroup, Misc, dives into a lot of disjunctive compatibility areas: file uploading, Flash, YouTube etc. Please do read the linked-in articles for more info if interested - here, I won't waste any time on telling the same stuff again. As can clearly be seen, Opera Mobile is the best of all in this group, particularly YouTube video-wise.

Speed, dynamic RAM memory usage benchmarks group: on Windows Mobile devices with, typically, heavily restricted CPU and memory resources, it’s very important Web browsers don’t tax neither the CPU nor the memory much. That is, they load the requested Web page as quickly as possible and try to radically reduce their memory consumption. As there are really radical differences between the different browsers, a Web browsing-related roundup MUST elaborate on these quantitive results.

Overall rendering speed: PPCMag test loading speed: in this test, I’ve measured how much time it did take to completely download and render the linked test page. Note that I’ve repeated the tests in different rendering modes to see what their effect on the overall rendering speed is. In general, I’ve made the tests on two current devices: the WM5 VGA 624 MHz Dell Axim x51v (running the A12 ROM) and the WM6 520 MHz VGA HTC Universal. In every case, I’ve noted which of the two I’ve measured a result on (the x51v is slightly faster, which is also reflected in the results).

Overall memory usage: program itself with a blank page (important particularly for HP iPAQ rx1950 / Palm Treo 700w users with ~11Mbytes of free RAM at most). Note that the PIE plug-ins show additional RAM usage, in addition to the "base" PIE RAM usage. : in this test, I’ve measured the memory usage of the applications without displaying any Web page (as displaying pages may dramatically increase the memory usage.) Note that, as with the next benchmarks, I’ve done separate QVGA and VGA tests; I used the HTC Wizard running WM5 as the QVGA test device. The reason for this is pretty simple: on VGA devices, Web browsers have the tendency of taking up more memory. As can be seen, Opera Mini and PIE are the most memory-friendly, followed by Thunderhawk and, then, Opera Mobile. Then follow the other browsers: NetFront and, finally, Minimo.

Note that, with PIE plug-ins (except for the hybrid Webby), I’ve measured the additional memory usage. That is, don’t think Spb Pocket Plus / PIEPlus only require 56k / 90k RAM memory; that is, that they greatly reduce memory load. It’s the additional memory usage, added to memory usage the “base” PIE.

An opened, new tab: unfortunately, not only the Web browsers themselves take up memory, but also the individual windows you open in them. This is especially true of PIE plug-ins, which, in effect, need to load a brand new instance of PIE into memory. This is why they, in general, consume at least an order of magnitude more memory (per window) than non-plugin-based, multiwindow-capable solutions (NetFront, Minimo, Opera Mobile, Opera Mini).

PPCMag test memory consumption: totally independent of the above-mentioned cases (how much memory the program itself / an additional tab take) is the memory taken up by the in-memory representation by actual Web pages you visit. This, in general, in cases, may be even an order of magnitude larger than the original size of the page – for example, (in this respect) worse browsers (most importantly, PIE) may take 7-8 Mbytes of the meager RAM to load a 600 kbyte Web page with some icons in there.

In this test, I’ve measured the memory consumption of all the tested browsers upon loading the above-introduced, 590 kbyte-big PPCMag test page. As can clearly be seen, there may be even two orders of magnitude differences in the results: while Opera Mini takes very little memory, PIE (the, in this respect, worst-behaving browser) takes between 7.5 and 9Mbytes.

Network connectivity group: in here, I’ve elaborated on generic network connectivity questions / issues.

Proxy support? If it does support proxies, does it require the proxy settings entered locally, or, does it get from the system-wide Connectivity framework?: Is the given app able to use proxy servers?

Proxy servers can be very handy in a lot of respects. Please see this article (also linked from this PPCMag article) on the usage of proxy servers. Also, you may want to read this article for more information on configuring proxies on the PPC/switching between them.

Opera Mobile and Minimo both support locally-set proxy servers.

As you can see, PIE, starting with Pocket PC 2002, uses the system-level proxy server setting. PIE plug-ins also use them as they have access to all the PIE resources. NetFront is also able to do the same, but you can also supply a different proxy server to it locally (which is the preferred and easiest solution in most cases). Thunderhawk and Opera Mini have no proxy support at all.

Proxy-based anonymity?:

If you use proxies, you can also anonymously surf the Web (please see this and this article on anonymity). This is why PIE (with all its plug-ins), Minimo, Opera Mobile and NetFront are preferred for anonymous surfing. TH, while it doesn’t support proxies, doesn’t pass anything client-related (no IP, no ThunderHawk username) to the HTTP server, so, it can safely be used for anonymous Web surfing too. Opera Mini, unfortunately, does pass the client IP in an extended HTTP header.

Does use the PPC2k2+ Connections framework to diff. between The Internet/Work connections?: You may have already run into the The Internet/Work distinction, which effectively plagues the life of a lot of people. PIE is based on this paradigm; this is why you run into a lot of ‘can’t connect’ messages because of just using the opposite type of connection of what’s needed.

Non-PIE browsers aren’t based on this framework, which is a big plus with them, at least for people that don’t understand the The Internet/Work distinction (it’s not easy stuff; furthermore, it’s not really documented either).

Bandwidth reduction: GZIP/Compress support? Does it really work?: HTTP browsers that support GZIP compression (please read this article on this subject) and support working through proxies (the case of Toonel – more on this later) may deliver a big win in bandwidth usage.

Toonel-compatibility?: Toonel is a great and, even better, free online HTTP compression service, a great friend of everybody not having unliminted (or very fast) Internet access. It requires explicit proxy support (and manual configuration) in the Web browser. In this row, I’ve noted the compliance of PPC Web browsers with Toonel. As can be seen, all of the "big" titles support Toonel because of the proxy support. It's only client-server solutions like Opera Mini, Thunderhawk (and DeepFish) that don't support Toonel.

Saving, downloading group:

Saving the current (Web) page (also see this)? (Note that it can even be a, say, as textual "rendered" CAB file too!): This shows if the browser is able to explicitly save Web pages. As can be seen, most of them do, Minimo, the two Operas and Thunderhawk being the exceptions. Some of the browsers (NetFront, PIEPlus, MultiIE) can even make a full save, downloading all the resources as the desktop IE in File/Save As - see the default Web Page, complete option in the Save as type: drop-down list.

Please note that the inability to explicitly save pages shouldn’t be a showstopper: you can get the Web pages from the cache of browsers that have local caches. It requires some manual work and searching, though. Consult the Download Bible for more information.

Save link directly to file, w/o opening? (""Save Target As...") (also see this): should you save something without actually peeking into it, you will want to look for browsers that do support this kind of functionality. (Please, as with the other rows in this group, do consult the Download Bible for more information on this subject - it's way more complicated than it seems!)

Co-working with HandyGet : Currently, HandyGet is the best Windows Mobile downloader tool/accelerator. In here, I’ve elaborated on whether it’s able to automatically “capture” the binary URL’s clicked in the browsers in order to download the file inside itself.

File download (NOT "Save Link Target"!)?: without relying on features like the above-mentioned "Save Link Target", is it possible to download files if they are offered for download (that is, if they are of binary content); is it possible to select a destination to store the downloaded file at. (Again, check the Download Bible for what this implies.)

Caching; cache benchmarks group: most Web browsers use local file stores called “caches” to store Web content locally, on the WM device. All this is in order to quickly speed up transfers and lower data usage. These caches, as they are stored in the file system, may result in a variety of problems, particularly when you visit pages with more than a handful linked resources (for example, images). In these cases, the sometimes vastly reduced file creation speed of non-RAM (read: flash ROM) media – for example, the built-in, default storage in all WM5+ devices. Please also see the related article What do you need to know about optimizing storage card speed? for more info on the speed issues caused by trying to write dozens of files to a flash ROM-based file system.

In here, in addition to elaborating on whether its size is settable, I’ve also elaborated (see Relocatable?) on whether the cache can be relocated to a storage card / RAMdisk etc. Note that, should you relocate it to an even slower medium (as are most of today’s non-high-end memory cards), the page loading times may become even worse with browsers (particularly sensitive to this problem is PIE), particularly when there are many files to store in the cache. In these cases, you will REALLY want to consider disabling caching entirely or using an area, RAMdisk, in the fast dynamic RAM (the program memory) to store the files. RAMdisks, however, have their share of problems (see the linked RAMdisk article).

I’ve benchmarked all the caching-enabled applications in separate scenarios. First, I’ve benchmarked them in my WM6 HTC Universal, using its built-in storage memory for the cache. Second, using a RAMdisk; third, using a VERY slow-to-write to, cheap SanDisk 1Gbyte SD card. As can be seen, with the latter card, PIE’s results are much worse than in the default or the RAMdisk one. Note that the results starting with + mean additional time needed for caching – in addition to the non-cached or the default case.

In Explicit cache navigation?, I’ve elaborated on whether it’s possible to examine the contents of the cache from inside the browser itself, as is the case with NetFront.

Finally, in Offline mode: Highlighting favorites present in cache (like on desktop browsers?) Loading cached pages without a connection? , I've elaborated on whether the browser supports showing what's available in the cache and what not. In the Favorites list, highlighting available pages is a pretty nice feature of all PIE’s except for WM5 (where, for some reason, it was removed). The second part of the test concerns cases of browsing without an internet connection, just from the file system cache. As can clearly be seen, this is not always possible.

Images group: in here, I've elaborated on image saving, (alternative) image text inquiring and wallpaper setting capabilities. As the latter (wallpaper setting) no longer works in any current Web browser or plug-in, you'll want to consult my well-known (Please read the "Today Wallpaper Bible" (alternatives: MobilitySite, AximSite, PPC Magazine, FirstLoox, BrightHand)) for more information on reusing downloaded / saved images as Today wallpapers, should you ever want to reuse an image on the Web as your wallpaper.

Copy/paste support group: I've elaborated on whether it's possible to directly copy a link to the clipboard and whether the browser supports arbitrary text selection from the given page.

As far as link copying is concerned, should it be missing with a particular Web browser / PIE plug-in, you can still do the same with just clicking the link and, then, when it's displayed in the Address bar, just stopping the loading (if you don't need to see it) of the page and copying the address from the Address bar to the clipboard.

As far as the second (text copying) is conerned, all browsers support it, except for Thunderhawk and Opera Mini (and the forthcoming DeepFish).

Hardware buttons not related to scrolling group: here, I've elaborated on hardware button assignment capabilities, which is REALLY useful and supported by some Web browsers (and PIE plug-ins). Assigned buttons can make operation (for example, the Back button) much easier, particularly if you don't like / can't use the touchscreen on a non-Smartphone (non-WM Standard) device. I've also elaborated on the WM5+ softkey support, which, traditionally, hasn’t been the strongest point of some browsers.

Scrolling group: you may want to prefer scrolling down/up the page (OR, select a link) using hardware keys (or the scroll wheel / jog dial or, alternatively, the SmartSKey-redefined volume slider, when available) instead of using the scrollbar (or, screen dragging) on the touch screen (if your device has a touchscreen at all). In these cases, you will most probably want to know what scrolling capabilities the given browser has and whether it's possible to override / change them.

In a nutshell, there are two traditional ways of scrolling: the "scroll one page at a time when you press the Up/Down arrows" ("page" scrolling) and "highlight the next link above/below/on the left/on the right when you press a directional key and scroll the screen contents by a page when there's no visible link in the given direction" ("link" scrolling). In addition, some browsers also offer the capability for "line" scrolling, which scrolls the screen line by line.

Traditionally, PIE in operating systems prior to WM5 utilized page scrolling and, starting with WM5, link scrolling by default. The switch to the new paradigm took place to make it possible for non-touchscreen-enabled smartphones to select (click) links to follow (and to let for one-handed operation even with touchscreen-enabled devices). However, the change to link scrolling wasn't really welcome by many users because it meant, sometimes, multiple, inconvenient, awkward keypresses to scroll down the screen contents.

There are a lot of different solutions to the problem, all of them explained / shown example screenshots of in the chart. Of them, hybrid solutions are the best and most usable. This is particularly true if you occasionally would like to use your otherwise touchscreen-enabled WM device in one-handed mode. Then, while still having the ability to both quickly scroll up/down the contents ("page" scroll), you also have the chance to do some link scrolling. This can happen with either the same keys (not) used with press-and-hold also used for page scrolling, or with different hardware facilities (either a scrolling wheel/jog dial or a redefined volume slider) to do the link scrolling.

As far as the first group (doing page/link scrolling with the same hardware facilities) is concerned, NetFront has an interesting scrolling behavior; with the brand new, 3.4 version of NetFront, you can fine-tune how the Up / Down keys behave; then, if you, otherwise, use link scrolling with the D-pad, you can still instruct NetFront to scroll through several pages up / down when you long-press (press and hold) the Up / Down key. (Note that the default behavior is immediately switching to the PagePilot mode for quick navigation.)

Also the scrolling model of Webby is of special interest: when you press the Down key, a page scroll will take; when you press Up, line scrolling. With this, you can still quickly scroll through a document without having to suffer from the disadvantages of link-only scrolling and, when you do need to access a link, you can scroll down one page and, then, gradually up (and left/right when there are several links in row) to get to a link. This is a very clever approach more closely modeling traditional user behavior.

Note that you are very lucky if you have a WM5 device with a real volume slider (for example, a HTC Universal, Wizard etc.); then, you can use one of the best, free tool meant for these kinds of devices, SmartSKey. With a redefined volume slider, you will always have page up/down scrolling in PIE (including all its plug-ins), (the new) Opera Mobile and NetFront (but, unfortunately, not in the other browsers); then, you can safely leave the D-pad in the default Line scrolling mode.

User-Agent group: the ability to redefine the so-called "User-Agent" can prove very useful because many Web sites check this information and act differently on mobile and desktop Web browsers. The ability to redefine this information can be very important because

1. many sites may refuse to provide (usable) content for a mobile browser introducing itself a mobile browser to the server, even when the client would be able to meaningfully render the contents. Just an example: while Opera Mobile's JavaScript and Iframe support is so darn good that it’s even able to make use of the very useful Gmail address autocomplete, Gmail switches to PDA view NOT offering autocomplete when it sees a mobile browser (including, by default, Opera Mobile too).
2. many other sites rely on for example authentication requiring a browser to identify itself as a desktop, while they aren't really using the advanced scripting or ActiveX capabilities of them.

In these both cases, redefining the User-Agent can prove very useful.

Note that you won't always want to redefine the User-Agent. There are many Web sites that, upon recognizing a mobile browser, provide mobile-/bandwidth-friendlier content. Just a few examples: the Smartphone & Pocket PC Magazine blogs, Pocket PC Thoughts, AximSite, FirstLoox etc. With these sites, it can prove very useful to be able to dynamically switch the browser identification (User-Agent) to the default (mobile) setting to get the mobile content.

Built-in browser identification change : in here, I've elaborated on whether the given browser / plug-in is able to change the User-Agent from inside the application.

On-the-fly external browser identification change visible without PIE restart in tabs opened after change? (Everything is +, also showing that all reviewed PIE plug-ins load a full copy of PIE into memory for each and every tab, unlike the old ftxPBrowser, which does require a full restart.) : As has already been pointed out, most PIE-based apps (except for ftxPBrowser) load an almost new copy of PIE into memory when a new browser tab is opened. This, on the other hand, also means that registry changes, which PIE only notices when it’s started, will also be visible after opening a new window (because PIE also reloads the registry), without even exiting PIE.

This can be of tremendous help. Let’s assume you prefer visiting a banking site pretending to be desktop browser (because the page just doesn’t let in, say, non-desktop-IE browsers), while you would like to access the, say, the PPCMag blog or Pocket PC Thoughts pretending to be a Pocket PC client so that you receive lightweight-formatted content. And, you would prefer doing this at the same time: in one window you browse online banking pages, in another one you browse the Pocket PC-optimized pages of the above-mentioned sites. It’s indeed possible if you always remember which tabs you opened after toggling the User-Agent.

Cross-posted to: PPCT, MobilitySite, AximSite, XDA-Developers (2), BrightHand, HowardForums, thesmallpicture, FirstLoox.