Well, the Gingerbread update finally became available, and I promptly installed it, hoping it would sort out all of my issues. Long story short: it didn’t. In fact, if anything, I kept getting an annoying message to the effect that the Tethering Manager (which I don’t use) was force-closing every time WiFi was turned off and then back on (for instance, every time the phone comes out of sleep mode, which happens, oh, maybe 20 times a day minimum?)

I decided to take matters into my own hands. With some trepidation, I went into the Settings menu, then Privacy, then Factory Data Reset… poof! All my apps, messages, data, etc. were gone. The phone was back in a pristine state—or, more precisely, in the pristine state of a Captivate running Gingerbread, as I did the reset after installing the OS update. After a very lengthy and frankly distressing wait, I had to reenter my Google credentials…

…and I was in phone heaven!!! True, I had to reinstall all my apps, but, since I use Google Contacts and Calendar, all my, well, contacts and appointments came back automagically. I really had no additional important data on my phone that was not backed up elsewhere. In particular, I didn’t care about SMS (text) messages, which do get wiped out. The huge payoff is that the phone is now fast. As in, it has never been this fast before! It almost feels like a new phone altogether. The browser just flies (this is probably also courtesy of Gingerbread of course), and Google+ displays the Stream almost instantly; I have yet to see a force close. Bottom line: the phone now officially has teh snappy. I can barely recognize my Cappy!

A few things I noticed:

• Android Market keeps track of the paid apps you have purchased, but forgets the free apps you have downloaded. Not a big deal, but you may have to remember what you installed if you didn’t pay for it
• When I did the factory reset, I didn’t actually wipe the internal storage as well. However, Google Music lost track of the songs I had synced on my phone. I had to do a bit of cleaning up afterwards, but again nothing was lost.
• If you use Google’s Two-step authentication (hint: you should), you need to reenter your application-specific password on your phone; if you didn’t save it, you must generate a new one (and delete the old one). Also, you need to redownload and reregister the Google Authenticator app.

Finally, one word of warning: if you do have important data on your Cappy, do back up!

Let me say, first of all, that I’m definitely not leaving the Mac platform! First of all, my office machine is a trusty 2008-vintage iMac that I have no intention of replacing, except with a newer iMac. Second, I still use my 2008 Macbook Pro, though rarely. Third, my wife uses a first-generation iPad and a 2010 Macbook Pro, and is very happy with both.

So why this, er, countercyclical (OK, contrarian) decision? Quite simply, I wanted a tablet that could double as a real, no-compromise, LaTeX-capable computer. Apple, sadly, does not make one. Yes, I know about the Modbook, basically a standard Macbook (Pro) retrofitted with a digitizer and tablet frame; it’s about twice as expensive, once you factor in the price of the Mac laptop, and weighs about three times as much as the Samsung 7 Slate — S7S henceforth.

I plan to post extensively on my experience with this machine. I certainly had to figure out ways to deal with the basic fact that Windows 7 and, more importantly, Windows apps (like Mac apps) are simply not designed for touch input at all (with a few notable exceptions). Let me start with a few thoughts, in no particular order.

Windows 8

The S7S is essentially identical to the developer machine that Microsoft handed out at the Build conference, where Windows 8 was first demoed in public. Microsoft has stated that any machine running Windows 7 today will be able to run Windows 8 as well, and the S7S is a very recent machine to boot. But, in addition, the S7S is one of the tablets that Windows 8, and Windows 8 apps, are being developed on! Thus, this machine will likely run Windows 8 well. Indeed, you can run the Windows 8 Developer Preview right now. Which brings me to my second observation…

The Tablet PC Review forum

If you own the S7S, you just have to follow the Samsung forum at Tablet PC Review. It is an amazing resource. Want to find out how to install and run Windows 8 on the S7S? Looking for cases that fit the still-new 11.6″ tablet form factor? Having issues with Samsung’s Easy (ha!) Software Manager? Want to check for potential issues before installing the latest BIOS update? It’s all there. People are extremely nice and helpful. The signal-to-noise ratio is very high.

Why not buy an iPad / Android tablet?

Because I don’t want to carry around two devices, when one will do just fine — actually, great, not fine. We already own an iPad, though I can’t really say I can use it: my wife and kids totally own it. That’s OK The point is, getting another tablet for media consumption purposes and web browsing seems like a luxury I can’t justify. Plus, it’s inconvenient: if you have and use two devices, you most likely will want to sync them (photos, music, etc.). This is no problem if you are at home and have WiFi, but we travel quite a bit. Until we have pervasive broadband access via the cellular network, at reasonable prices, worldwide, syncing will be a pain (think iTunes). If your laptop is also your tablet, there’s no syncing involved.

Why not a Macbook Air?

This is a tough one. As far as portability is concerned, the MBA is every bit as good as the S7S, or any tablet for that matter. It is a (small) laptop, which means that you can work comfortably while holding it, well, in your lap. In this respect, it is definitely superior to a tablet, even if the tablet comes with a case stand or dock, and a keyboard, like the S7S does.

But the point is that you really don’t want a keyboard when you are reading a book using the Kindle app (or similar), and arguably even when you’re casually surfing the Web, or triaging email. By the way, the S7S comes with a beta of the Swype keyboard for Windows. It is simply amazing; I use it on my Android phone, but it is even better on a tablet. Hint: make sure you read the advanced tips on the Swype Web site.

The S7S provides many of the form-factor advantages of an iPad or Android tablet, but is a full-featured, relatively powerful PC (Core i5 processor, 128GB of SSD storage, 4GB of RAM, a decent Intel graphics processor). You can get real work done on this machine, no matter what your line of business.

By the way, this thing is fast. Try surfing the Web on the S7S and on an iPad side by side. There’s no comparison. And, yes, of course Flash runs just fine on the S7S, and while I loathe it, certain sites (fancy restaurants being the offenders I discovered more recently) just insist on using it.

Not for everyone

Lastly, I should be very clear about this: the S7S is not for everyone. If you want a pure consumption device, get an iPad. It lasts much longer on a charge, and it is designed for touch from the ground up. If you don’t care about the tablet form factor at all, get a laptop. Most Windows apps require adaptability and, above all, patience if you are using your fingers instead of a mouse or touchpad.

Watch this space for tips, tricks, and adventures in the world of Windows tablets!

As usual, you will get the updated plugin automatically if you installed LaTeXTools using Package Control, which, as I noted earlier, I strongly encourage you to do.

Switching to the PDF viewer after compiling

By default, LaTeXTools keeps the focus on the Sublime Text 2 (ST2) window after compiling a TeX source file to PDF. This is convenient in two scenarios. First, if you have a large screen (or two monitors), you can keep ST2 and your PDF viewer side by side, and just glance at the output to see that all is OK. It would be quite annoying if the viewer window was brought to the foreground in this case: in order to continue editing, you would have to manually switch back to ST2 (using Alt-Tab or Cmd-Tab, depending on your platform). The other scenario is when you are making many small changes to the file sequentially; you compile to make sure that there are no errors or warnings, but do not need to check the PDF output every time. I also have friends and coauthors who simply don’t need to look at the PDF output all that often–they can read LaTeX easily, and would much rather not be distracted by the viewer window popping up.

That said, it is sometimes convenient to switch to the PDF output, especially if you are using a small screen and running both ST2 and your viewer in full-screen mode. In such cases, it would be nice if LaTeXTools could automatically bring up the viewer after compiling. Yet, the previous paragraph gives a few reasons why this shouldn’t be hard-wired.

Enter the Toggle Focus command (bound to Shift+Win+F on Windows and Ctrl+Cmd+F on OSX). It does what you think: every time you invoke it, it changes what the Build command does after compilation. Again, by default, the Build command refreshes the PDF viewer but makes sure that ST2 keeps the focus; so, if you invoke the Toggle Focus command, the next time around Build will actually tell the PDF viewer to grab the focus, i.e. pop us as the frontmost window. Invoke Toggle Focus again, and you get the default behavior back. Every time you switch, a short notification appears in the Status Bar (at the bottom of the ST2 window), so you know what you just did.

Note that this setting is preserved with your session; if you quit ST2 without first closing the tab you are working on, ST2 will remember the status of the focus toggle.

Wrapping existing text in LaTeX commands or environments

The current facilities for entering LaTeX commands or environments (Alt-Shift-[ and Alt-Shift-] on Windows, Cmd-Shift-[ and Cmd-Shift-] on OSX) are useful when you want to first specify the type of environment or command you want, and then enter text in it. However, sometimes you want to wrap some existing text in a command or environment. The most common use case is to emphasize text, or make it bold. I don’t use this feature that often myself, but a colleague mentioned it as one of the main reason why he was sticking with TextMate for the time being.

I have now added a nice collection of wrap commands. They are all bound to Alt-Shift-W on Windows and Option-Shift-W on OSX, followed by an additional key (i.e. you use a “key chord” to invoke them). You must select some text prior to invoking these commands.

Alt+Shift+W followed by n wraps the selected text in an environment, which by default is called “env”: that is, if “blah” is the text currently selected, it gets replaced by

\begin{env}
blah
\end{env}

and the env is highlighted. Enter your desired environment (e.g. “theorem”). When you are done, hit Tab to jump to the end of the environment.

Alt+Shift+W followed by c instead enters a command: “blah” becomes \cmd{blah}, with cmd selected so you can change it to whatever you like.

Finally, a few common commands have dedicated key bindings: Alt+Shift+W followed by e, b and u respectively give you \emph{...}, \textbf{...} and \underline{...}.

Happy TeXing!

Now for the fixes. In no particular order:

• I expanded the README file so it now covers most, if not all, of the current functionality. I also provide some ideas on how to troubleshoot issues. By the way, if you do find something is wrong, please let me know!
• I finally understood that latexmk does not automatically force the TeX engine (e.g. pdflatex) to work in batch mode. Errors will cause latexmk to stop. You can check this by running it from the command line. However, on OS X, somehow invoking latexmk via subprocess.Popen still works, in the sense that all processes terminate regularly and the plugin can then look for errors in the TeX log file. On the other hand, on Windows, using TeXlive, this is not the case: the build command essentially hangs, and even if you terminate it, the pdflatex process keeps waiting for input in the background. This makes further compilations fail (because the synctex file is locked), and forces you to terminate the stray process with Task Manager. Ugly! Well, I hope I have fixed at least the most egregious causes of such inappropriate behavior by passing the -silent option to latexmk.
• A user had a lot of trouble with on-the-fly conversion of EPS files to PDF on OS X, using the epstopdf package. Ends up the problem was path-related: epstopdf relies on Ghostscript, and MacTeX puts Ghostscript in /usr/local/bin rather than /usr/texbin. When compiling from the command line, everything worked–because MacTeX also helpfully fixes your $PATH; but, building from Sublime Text 2 always failed. Adding the former path in LaTeX.sublime-build fixed the issue. Thanks Jon and Sublimator for helping me track down this bug!
• SumatraPDF, the supported PDF reader on Windows, only reloads recently changed files if they are local: it does not automatically reload files on network shares when they change. As a result, recompiling a file on a network share resulted in SumatraPDF not automatically showing the updated text. This was a problem for me when running Windows in a Parallels virtual machine: Mac directories are accessible are network drives, and there is no problem editing files with Sublime Text 2 on the virtual Windows side; however, I had to manually reload PDF files every time. Bottom line: I now explicitly force a reload before issuing the forward search DDE command.

Enjoy, and as usual report back any issues you may have!

Mail servers: POP vs. IMAP

As you may know, there are two main ways, or protocols, to access your mail on a server: POP and IMAP. POP (“Post Office Protocol”) is a simple, older mechanism that allows a mail client (Apple Mail, Thunderbird, Outlook, etc.) to ask a mail server whether you have new mail, and download messages to your computer or connected device. You can also tell the server to keep a copy of the messages you download, or delete them. So, the basic premise behind POP (as I see it) is that the mail server is just a way station between the machine sending mail to you, and the mail client on your PC.

POP makes sense in a world in which Internet connectivity is slow and storage space on the server is expensive. In other words, not in today’s hyperconnected, always-online world. The basic premise behind IMAP, on the other hand, is that your mail resides on the server; your mail client can access it at any time, of course delete it if you wish, and perhaps even cache it locally for performance reasons. However, the “official” repository of your mail is on the server.

A huge advantage is the fact that it becomes very easy and safe to access your mail from many different machines—your PC at work, your laptop at home, your smartphone, etc. The mail client on each device gets to read mail on the server, but there is one central repository. With POP, you have to make sure that you instruct your client to “leave mail on the server”: if you don’t, the day will come when you are out of the office, need to read an email your boss sent you the day before, and realize that you already downloaded that email from your work PC and deleted it from the server.

IMAP has all kinds of other advantages, such as folders. The feature I want to focus on today, however, is flagging.

Custom server-side flags

The IMAP mail protocol supports flagging a message as important, and of course Mail supports this. The neat thing is that flagging is a server-side attribute: it is stored on the IMAP server, not on the client. Thus, if you flag a message on one machine (say, your work PC), then later retrieve the same message on another machine (e.g. your laptop, or your smartphone), the flag is there.

However, the IMAP protocol also supports “keywords”—additional status indications whose interpretation and visual rendering is left to the individual mail client, but that are also stored on the server. Older versions of the Mozilla Thunderbird mail client used IMAP keywords to store “labels”, which allowed you to mark a message as “Important”, “To do” or similar; messages decorated with different labels were also displayed different (non-black) color. More recent versions have “tags”, which work similarly but can also be customized by the user.

Well, it turns out that, when you flag a message in Lion Mail, you can choose a flag color other than the default red, and the color is stored on the server. In the Lion Mail client, notice that there is a pull-down menu (a triangle pointing downward) next to the Flag button: that’s where you choose the color of the flag you want. If you access your mail from a different mail client, you don’t see the color, but you do see the flag: that is, things degrade gracefully. Yay!

IMAP Access to Yahoo mail

Well, the title says it all. Yahoo offers free POP access to its mail server, in addition to access via its Web-based interface; if you want IMAP access, you have to sign up for premium service. The only exception is the iPhone and the iPad: Apple must have a special deal with them, and iOS devices do get access to Yahoo’s IMAP server.

My wife has had a Yahoo account for as long as I’ve known her, and she is very attached to it. Despite my repeated attempts to get her to switch to Gmail, she remains faithful to Yahoo. Thus, up until OS X Lion, she accessed Yahoo via their POP server on her laptop, and via the special IMAP server on her iPhone and our family iPad. This is an unholy combination; although we always tried to be careful, between laptop upgrades and mail client updates, I have always been worried about inadvertently unchecking the “leave mail on server” option, thereby making my wife’s email inaccessible from her iPhone and iPad—nowadays her primary mail devices by far.

Furthermore, my wife tends to have a rather large Inbox. Actually, this is not surprising if you are used to POP: a mail client can offer to store messages in a folder on your PC, but the server knows nothing about this. So, if you want to have cross-device access to your mail, and still keep old messages for a while, well, the only way is to leave them on the POP server, so that they show up in your Inbox everywhere. Indeed, iOS does a good job in this scenario: you only get to see 50 or so messages at a time, but if you need access to older messages, you can usually get to them quickly via search.

The problem with this approach is that, if iDevices are your primary mail machines (via IMAP), the rare time you access your mail via POP, your client has to download a lot of messages. For us, this got so out of control that my wife simply stopped using her laptop for email. This was a problem if she needed to print an email: the only practical way for her to do so was to forward the message to me, and I’d print it for her.

Well, no more: as of OS X Lion, Apple Mail has full access to Yahoo’s IMAP server. My wife is ecstatic: not only does Mail look very much like the iPad mail client, she also sees exactly the same messages with no trouble at all, and no time spent downloading old messages!

If you are a Yahoo mail user, this is a big win! A hint: if you have already configured access to Yahoo via POP, the easiest thing to do is to (archive your messages and) delete that account, then use Apple’s configuration wizard to add Yahoo back. Go to System Preferences (notMail’s preferences!), choose “Mail, Contacts and Calendars”, and click on Yahoo. Enjoy!

10/13/11 3:46:14.887 PM com.apple.launchd:
(com.apple.Kerberos.kdc)
Throttling respawn: Will start in 10 seconds

(I broke this up to make WordPress happy; in reality the above is a single line in the log file).

Basically, launchd was trying to run a Kerberos (i.e. authentication) related process, and repeatedly failing. I am pretty sure that this started happening after upgrading to Lion.

Anyway, there is a quick fix, thanks to this post by Craig Newell in the Apple support forums. From the Terminal, issue the following command:

sudo /usr/libexec/configureLocalKDC

(you then have to enter your password). This fixed things for me. See the aforelinked post for a brief explanation.

And here’s the same snippet, as rendered by Safari 5.1:

Another example, this time from a Wikipedia page I reference below: today’s Chrome…

…and Safari:

Until last night, I could not spot any difference in the rendering (or, if there were differences, they were not so jarring to my eye). Now, there are two key differences.

At the “macro” level, fonts as rendered by Chrome look thinner. Subjectively, they are not as legible. They also look slightly washed out.

At the “micro” level, if you magnify the Chrome and Safari snippets a few times, you will see that Safari uses subpixel rendering: loosely speaking, this means that the “edges” of character shapes (glyphs) are actually drawn in different colors, not just different tones of grays. On the other hand, Chrome seems to have inexplicably switched to purely grayscale rendering of fonts. [Note: the difference is even starker in the original TIFF screen captures, but you should be able to notice in the WordPress-friendly JPEG pictures posted above.]

The “micro” view explains the “macro” perception. Subpixel rendering is used precisely as a way to increase the perceived resolution, and—if used tastefully, as it is on Mac OS X—to achieve greater typographic fidelity. For whatever reason, today’s Chrome dev build seems to have switched that off.

Please, Google, bring subpixel-rendered fonts back, or I’ll have to switch to Safari. You already took away the ability to move between bookmark menus; please don’t take away my nice fonts!

If you were using SumatraPDF with another text editor that supported inverse search, you probably didn’t have any trouble reconfiguring inverse search to work with Sublime Text. However, it turns out that a fresh install of SumatraPDF does not show a GUI for the inverse-search command, until you actually open a PDF file that has actual synchronization information in an accompanying .synctex.gz file.

The revised README file shows how to proceed. Basically, if you already have a PDF file with an attendant .synctex.gz file, open the PDF in SumatraPDF, and you’ll see the GUI when you select Settings|Options. Otherwise, you can either compile any latex document with pdflatex and sync information (i.e. compile with pdflatex -synctex=1 ), or else configure SumatraPDF from the command line. See the README file for details. As usual, you can read that file at Github (you may need to scroll down a bit).

I’m sorry things are not more straightforward. Good luck!

Also, I finally got around to beefing up the README file for the plugin. It’s in Markdown format, so it’s human-readable, but you can see it in all its HTML-formatted glory if you just go to the LaTeXTools page at Github and scroll down. I hope this will encourage more people to give Sublime Text 2 and LaTeXTools a whirl.

One final note on the choice of TeXlive vs. MikTeX. As I see it, the trade-off is this: MikTeX’s management tools, and its DVI previewer Yap, are Windows-native, whereas TeXlive’s management tools feel like ports of Unix tools, which they are (even the GUI ones). On the other hand, TeXlive has full support for spaces in file names and paths, which MikTeX currently lacks. As a further advantage, TeXlive has a working version of latexmk, which means that it is very easy to change the tex engine from pdflatex to, say, xelatex: just change the appropriate line in LaTeX.sublime-build. Finally, TeXlive on Windows is essentially the same as, and package-wise in sync with, MacTeX on OSX, which may be convenient if you’re a platform hopper like me.