Bvckup blog

Master Feature List

Wed, 13 Jun 2012 10:35:00 -0700

Master Feature List - a comprehensive list of features getting done in V2 is now available.

If you are subscribed to the updates, you will be receiving an email shortly with a link that lets you vote features up and down.

Would you look at that... an update!

Wed, 23 May 2012 11:45:15 -0700

Check it out here - http://bvckup2.com/news/23-05-2012

Moved

Fri, 30 Mar 2012 16:12:24 -0700

Bvckup and Bvckup2 sites have been moved to a new server. The update.bvckup.com has been moved to the same server as well. Should you notice any 404s, empty looking pages, strange messages or something equally amusing, do let me know.

On Backups

Thu, 01 Mar 2012 18:19:51 -0800

A classification of data backups, and how conventional backups (do not) really fit in.

Check it out - http://swapped.cc/on-backups

Merry Christmas!

Sat, 24 Dec 2011 15:39:00 -0800

Version numbers don't matter

Wed, 14 Sep 2011 16:15:00 -0700

In the spirit of documenting changes that are obvious, but only in retrospect I want to go over a new versioning scheme that I am adopting for Bvckup. Its executive summary is -

No more dotted version numbers.

In other words, “Bvckup 2.1.3.14” becomes “Bvckup 2, revision 37” …or thereabouts.

In short

This is not an isolated change, but rather a part of a larger rework aimed at streamlining the user experience with software updates. The rework that includes three core changes:

Dotted version numbers such as 2.3.17 are replaced with a single revision number.
When prompting for an update, the list of changes will always be shown relative to the user’s installed version.
When prompting for an update, the program will also indicate the importance of updating based on what changes are missing from the user’s version.

The goal is to evolve from amorphous and pseudo-informative

to a more concise and actionable

The rationale is simple and it is laid out below.

Rationale

Dotted versions are a de-facto standard. Comprising major, intermediate, minor and sometimes “build” numbers, they are routinely used to indicate the type of changes that a software release carries.

The exact versioning scheme varies from developer to developer, and sometimes it’s different even between projects of the same origin… Ask me how I know :)

The problem, of course, is that trying to communicate the release summary through a string of numbers is a really odd thing to do. It opens things to interpretation and leads to critical updates being glanced over, useful changes mistaken for unwanted new features, and so on.

It’s also safe to assume that a vast majority of users don’t care in the slightest which specific version they’re running. They care if their version is the latest and what they’re missing if it isn’t.

This makes the version number all but useless, or as some would put it - meaningless. Without a change log, the number cannot be used for anything but understanding if the installation is up-to-date.

For this reason, version numbers need not and should not be elaborate. And for the exact same reason, the change logs need to go well beyond being just a laundry list of assorted changes. More on this below.

Simpler versions

There are several obvious alternatives to the dotted versions. These include:

A single number, incremented with each public release
A release date, i.e. we could just tag each version with the date of its release.

The latter does not accommodate releasing more than one update on the same day - which does happen when someone gets sloppy - but it can be easily remedied by:

A release date with an optional release number that is shown only when needed.
A month/year combined with a release number, e.g. “Sept. 2011, release 15.”

Mixing the date into a version provides an at-a-glance evaluation of a version’s age. But it also comes with the added hassle of localization. What June is for some, is Июнь for others. Just one of those challenges the support department lives for :)

In my case, I’ve decided to go with a release number only.

Update notifications

Not everyone likes automatic updates, and not everyone uses them. And among these people, there are few who update right away even if a program prompts for it.

There are plenty of reasons for that - some users are just busy, and some are generally cautious about (not) touching a working setup, the ever popular don’t fix what’s not broken attitude.

I would argue that by keeping update notifications concise and relevant and by allowing users to defer the decision, it should be possible to convince more people to upgrade, either right away or eventually.

1. Be concise

Remove everything that does not contribute to helping a user make an update decision.

For example, a new version number is largely irrelevant: it does not add anything, provided we explain the changes elsewhere. Remove it.

Some update notifications display both new and current version numbers, as if this helps with the upgrade decision. It does not. Remove both.

2. Be relevant

Have a list of changes ready. There is a very good chance someone who opted out of automatic updates will want to look at the changes.

Look at the user’s current version and only include changes that are relevant to the user’s case. A change log reaching back to the first pre-alpha release is useless. And so is the log that is relative to the preceding version.

It’s not about what changed; it’s about what’s missing.

Secondly, have all changes sorted by their type first - bug fixes, new features, smaller tweaks, etc. - and then further sorted by their importance. Not chronologically. Proper structure greatly simplifies the digestion of information.

Lastly, exclude irrelevant changes. These are the fixes for issues that were introduced and resolved in intermediate releases. For example, if the user missed two updates and the second fixed a problem caused by the first one, then such a fix is of no importance to the user. Hide it.

The result would be going from something like this -

to this (assuming the user is on version 1.0) -

Note the grouping and re-ordering. Note the lack of 1.0 and 1.2.1 changes and the lack of version numbers. Also note how important bug fixes appear in bold. Had there been critical fixes, they would’ve appeared in red.

3. Be helpful

Make a suggestion. We have a list of changes, and we know which ones are more important than others. So look at the most important one and tell the user if the update is required, recommended, or optional.

Required is for critical updates, i.e. “You’ll be sorry if you don’t update.”
Recommended is for new features, non-critical bugfixes, performance tweaks, etc.
Optional is for irrelevant fixes and smaller items such as adding a UI translation.

4. Be considerate

Give the user the option of making the decision later. Ask again in a bit and be intelligent about when. Interrupting a user who’s wiggling the mouse and typing on the keyboard is not wise; it’s annoying.

The best time would be after the machine has been idle for a while, typically indicating that the user has left for the day. When he or she comes back, it’ll be a new day and easier to sneak in a smaller decision such as making a software update before the routine starts.

Trivial to automate

On the development side, the process is easy to automate. When a developer finishes making a change, he enters its description, its type - a bug fix, a tweak, or a new feature - and selects if the change warrants requiring or recommending an update.

If the change reverses an older change, it can be tagged with a respective release number. This tag would then be used to filter irrelevant changes as noted above.

Lastly, a change may be assigned a relative weight to help position it properly on a sorted list of changes of its type.

Comments?

There is a thread on Hacker News for that.

Please confirm your email address

Thu, 18 Aug 2011 17:27:00 -0700

Just a quick post to share something that seems obvious in retrospect, but took a while to notice and fix.

A mailing list for those wanting to follow Bvckup and Bvckup 2 progress has been up and active over a year now. Subscribing to the list triggers a typical confirmation email with a link that needs to be clicked on. The odd thing was that a good half of those subscribed never bothered to do that.

So last month, after receiving a small rush of sign-ups from betali.st I decided to put things in order and run a bit of A/B testing. Good thing I did.

Simply changing the email subject from this

to this

increased the number of completed subscriptions from 51% to 82%.

The original subject read like… well… a confirmation rather than something that requested an action and people simply did not bother even opening an email. A simple but costly oversight that took me almost two years to realize.

Comments

There is a thread for that:

Hacker News
Reddit
Habrhabr (in Russian)

BetaLi.st

Thu, 21 Jul 2011 10:52:28 -0700

Made it to BetaList yesterday and wow, just wow. Steady thick stream of traffic and trending up on their weekly list of populars too.

Thanks for the sign-ups, mentions and tweets, guys! Very much appreciated. And a big round of applause to Mr. Köhlbrugge for conceiving such a useful site.

http://betali.st/startups/bvckup-2

Technical overview of Bvckup 2

Fri, 15 Jul 2011 13:45:00 -0700

I described several technical aspects of Bvckup 2 while it was on reddit.

As the development moves forward I will be posting detailed performance numbers for each listed feature, but meanwhile take a look at the overview to get a general feel for the amount of optimization finding its way into V2.

So here it is - A technical overview of Bvckup 2

Wed, 29 Jun 2011 14:47:00 -0700

Two-day sponsored post on reddit is now live - http://www.reddit.com/comments/iarao

About one quarter through and there is 2300 unique impressions so far, about 40 click-throughs, plus a bunch of opinions which is what I am after. All in all this seems to be beating Adwords hands down.

Bvckup 2, the Teaser

Mon, 20 Jun 2011 15:01:00 -0700

Things are moving forward and today I am very happy to announce the launch of the Bvckup 2 (teaser) website - http://www.bvckup2.com.

With this launch we are now starting to showcase minimalistic aesthetics of Bvckup 2 and introduce our visitors to the concept of refined single-purpose software.

This project is a bit of a gamble… or an experiment, if you will. I really want to try and gauge interest in Windows software that picks the design and engineering quality over the (over-)abundance of features and toolbar buttons. If such software indeed has its place in the Windows ecosystem, there’s quite a few clever ideas just waiting to be executed.

Interesting times ahead, stay tuned.

Teaser

Thu, 28 Apr 2011 00:45:00 -0700

Working on many things including a post on core optimizations and an update of the product identity. With latter trying to tone down some of the grandness of the current logo, to make it a bit more approachable and to better express the nature of product — a simple software that does just one thing, but does it really well.

Version 2 in broad strokes

Mon, 28 Feb 2011 15:57:00 -0800

Bvckup entered its public beta in July ‘09, and the past year and a half provided a good insight into what was working, what was missing and what needed fixing. And so it is now ready for its next major revision — Version 2.

The new and improved Bvckup will focus on three things - core improvements, the user interface and speed optimizations. From mundane but useful, to pretty, to nerdy-cool, it got it all.

Core improvements

These are notable changes of a general nature — useful, but somewhat boring.

Enhanced NTFS support — alternate streams, junctions, symbolic links, etc.
64 bit version
Optional archival of deleted items as per these comments
Further reducing the size of the executable and its run-time memory footprint

Two other features that I am considering are run as a service and versioning (either in a form of backup snapshots or per-file archives), but these are fairly big features and each may warrant a separate release.

User interface

This is where things start to get interesting.

New user interaction model whereby the configuration of certain backup parameters will be delayed until such parameters are actually used.

For example, if the destination does not support symbolic links or if a symbolic link points outside of the source directory, the app will ask what to do when it runs into these conditions.

This change is meant to streamline the initial setup of a backup job by not requiring a decision about what to do in exotic cases.
Updated the main window to display some status and progress details directly in the list of backup jobs instead of keeping them in the Information panel (that is also getting redesigned for further clarity).
Redesigned the file filtering interface, i.e. the file/directory inclusion/exclusion.
Optional backup job grouping as per these comments.
Advanced section in Preferences to allow for easy access to all .ini settings.

Speed optimizations

And this is where things are starting to get exciting… in a nerdy kind of way :)

Faster disk scanning, in part due to observing exclusion/inclusion filters.
Logic for minimizing or avoiding disk scanning whenever it’s possible.
File and directory move/rename detection.
Cache-aware reading — speeds up file reading from the disk cache if possible and falling back to direct disk access otherwise.
Asynchronous pipelined IO — evens out the load that IO activity puts on the OS and eliminates idling time.

Pipelined IO

Pipelined IO is the change in version 2 as far as speed improvements are concerned. It is a notch up from the “multithreaded” disk access favored by some other backup apps.

This IO model is substantially more complex to implement, but it results in a faster file access with lower system overhead. It also naturally supports various clever auto-tuning techniques. This is very cool, this is what gives the app its smarts.

Schedule (in very broad strokes)

Most of the speed-related changes are already prototyped and tested, what is left is the integration into the production code. The UI changes on the other hand are in the early stages of sketching, and this is what I will be tackling in the next few weeks (follow the Dribbble feed for the updates).

There is also a separate task of trimming the fat off the application, both its run-time resource usage and the executable size. This translates into another few weeks of work.

The website refresh is another big item, so all in all it’s going to be 2-3 months at least, and more likely double that. Good things take good time :)

It's alive!

Wed, 16 Feb 2011 10:55:00 -0800

This is to officially open this blog for business and to publicly commit to not spending any more time on tweaking its design. A month and a half of it should be plenty enough :)

Do take a look at a beautiful footer for the ways to keep up with the blog.

How Bvckup got its name

Sun, 13 Feb 2011 17:23:00 -0800

Bvckup started its life as a small copying utility that I wrote for my own use and back then it was going under the name of JustCopy. Despite its spartan interface and hard-coded configuration that early version proved to be useful. It was a solid prototype of a real product, and a real product needed a name.

Read on to learn how the app got its name and the logo.

Name is the first impression

Naming a product is hard, and naming it well is really hard. Name is frequently the first thing someone learns about the product, and this first impression - it really lasts, and it’d better be absolutely spot on.

Consider, an example:

I can call it a dinner spoon or I can call it a “soup delivering apparatus №2”. Which one sets a better tone for introducing a one-of-a-kind designer product?

Exactly :)

Making it memorable

Good names generally have strong concepts behind them. A concept may be as effective at making the name stick in a memory as hammering it into people’s heads with brand awareness campaigns. Repetition surely works, but so does the originality.

μTorrent for a really light BitTorrent client - that’s a very good name. But Voldemort for a distributed storage — this is a raw genius.

Examples are aplenty, and yet the product naming remains more of an art form rather than anything else. Its practicing artisans wield exotic tools like Swahili dictionaries and Scrabble tiles and not shying away from charging towards 10-20K per project.

As much as I wanted to try this route, it was clearly an overkill for a file copying tool.

DIY naming

The first step in the naming endeavor is to write down a bunch of words describing product’s core function and sprinkle with adjectives describing its attributes. Then stir and combine, see what comes out and go with the associations.

Here’s how it worked for Bvckup.

Function:

backup, copy, clone, replicate, mirror, sync

Attribute:

light, small, simple, intelligent, fast

With these as a starting point, started mixing and after a while arrived at the first option:

Small › Nano › Micro (courtesy of uTorrent) › uBackup › Buckup

Buckup was a complete LOL, of course, and yet it had something. It was short, it was surely memorable and it had a direct link to the nature of the program — something that is used to prevent data #uck-ups.

Set it aside and continued. The second take was through the simplicity angle:

Simple › Trivial › Vulgar › Backupus Vulgaris › Bvlgaris › Bvckup

Now this was interesting. Not as rude as the first option, but as short and as closely related to the backup word.

It also had neat typographic properties. V is a flipped over A and it has a basic form of a shield. V is also pointy so it can be shaped into an arrow, and arrows are good, they capture the motion and agility aspect.

Validating the idea

It was time to run the idea by other people, see if it holds up.

For that it really helps to know places where people are not shying away from giving real critique. One such place is a stable and thoughtful community over at Typophile forums, and the Corporate ID Design forum specifically.

The Bvckup thread is at http://typophile.com/node/54931 and the responses show that the #uck-up reference was still a bit too strong. That however was something that could be offset by means of visual design.

Wordmarks

I am a very big proponent of simple logotypes, and best among these are wordmarks. Take the name, set in a standard font and then tweak it to achieve a distinct look and desired feel.

For example, here is Landor’s, the branding agency behind lots and lots of identity work:

Or this:

Still unconvinced? :) Head over to David Pache’s excellent collection of 100 Brands of Interest and see how many of the brands are wordmarks. Parts two and three are here and here respectively.

In any case, back to Earth.

Developing the idea

With V being a focus element of the name, it was only natural to build the customization around it. The pointy shape of the V suggested an arrow and a bit of doodling on a paper yielded these two:

Further varying this and that in both designs led to a larger set of options:

Now it was just a matter of selecting a symbol that matched the target feel of the logo the best. This involved just sitting there, staring at the screen and subjectively sorting designs by their appropriateness for the task.

Anything with the sharp pointy angles was a no-go. Asymmetrical designs were inferior to those with symmetry, and smooth curves looked better than straight and jointed lines.

The result

After a day or so of intense staring the option at the top left appeared to be the best choice and after a bit more tweaking to smooth its outer angles, the mark was done.

It’s a V and a flipped over A. It’s a shield and it’s an arrow referring to stashing or bolting things down. It is also a very simple and distinct symbol in its own right that works nicely as an application icon.

Almost a year after someone pointed out that V doubled as a flipped over Greek letter Δ. Delta reference is of some significance because it’s the delta copying what gives Bvckup its mind-boggling speed.

In any case, there you have it - from a list of nouns and adjectives to the name and from your ol’ Gotham to the product logo. From start to finish - in about one month.

Finishing touches

Fri, 11 Feb 2011 23:29:00 -0800

Tying up some loose ends.

A gentle warning for those running NoScript plugin. Here is a demo.

A notice to introduce this blog on the homepage.

Regarding the timeframe

Thu, 10 Feb 2011 10:16:00 -0800

A very reasonable question arrived today over at Twitter:

ckasprzak @apankrat so is it a web design or app UI? Any ETA on v2?

And an answer to this question is this:

A guy brings a bolt of fine cloth to his Jewish tailor and asks him to make a pair of pants.

A week later he returns, but the pants aren’t ready. Two weeks later he returns, but the pants aren’t ready. Three weeks, four weeks, five weeks - no pants! Finally, after six weeks, the pants are ready, and they fit perfectly.

Nevertheless, the man can’t resist a jibe at the tailor. “You know,” he says, “it took God only six days to make the world. But it took you six weeks to make a single pair of pants.” “Ah,” says the tailor, “But look at this pair of pants … and look at the world!”

In other words, that screenshot is just a blog footer :)

The app UI and the main website designs are to follow. Some parts are sketched, but more work is needed. Tentative v2 development and its release schedule is due soon as well, and so is the list of v2 features.

Stay tuned, good things take good time, but they are coming.

Input control styling

Fri, 04 Feb 2011 15:21:00 -0800

Just started on slicing the blog footer comp, should be done by Monday.

The last thing before that was a matter of input control styling. After trying this and that the following was a set of lucky finalists:

As one can plainly see, the issue here is not the lack of choices, but rather their abundance. One good option typically leads to few more. When it rains, it pours kind of situation.

Single-input form

While the control looks nice, it also works in a fairly unique way.

Click on a control, type the email address and hit Enter (or click on the arrow), see it disable any further input and submit the email to the server. When the server responds, the control shows ✓ for a confirmation and ! for an error:

This is as compact as it gets without loosing usability. Consider it my humble contribution to the modern UX design :)

There is a live version on the existing Follow page. Feel free to play with it and I will clean up the junk submissions later.

The dots

These spinning dots… there’s something hypnotizing about them and it makes me want to put an artificial delay in the processing to keep them spinning a little longer. Doesn’t work with three dots, nor does it work with one (off-center) dot. Go figure.

Styling the blog footer

Tue, 01 Feb 2011 00:07:00 -0800

Half way through redoing the blog footer from its temporary form -

to something a bit more elaborated -

The footer will then further be recycled for the main website, so this ties into the styling work of the main site.

On sidebar styling

Fri, 28 Jan 2011 11:50:00 -0800

A pair of sketches to show a bit of the “behind the scenes” design routine. In this case it’s the styling of the sidebar information that accompanies every post.

The first option used icons as link labels:

It looked good on its own, but when considered in the context of the whole page it happened to add an unnecessary visual noise. Sidebar is a supplementary information after all and it should not be attracting more attention than the post itself.

The second option was much simpler and it relied purely on a typographic styling:

The issue here was of the implementation nature. The italics are Georgia and it does not render well on Windows in smaller font sizes. For this reason the labels had to be pre-rendered into the div’s background.

Icons

Had good time making the icons though :) Here’s the full set as seen on Dribbble:

Perhaps will have a chance to recycle them for some other project one day.