Ask HN: Please review my weekend side-project. | Hacker News Reader

Ask HN: Please review my weekend side-project.

whatkindoffileisthis.com

20 pointsabstractbill27 comments

This is essentially an online version of the unix 'file' utility that I hacked together quickly over the weekend (it actually uses 'file' on the backend).

There are many obvious directions I could potentially take this - for example improving recognition accuracy, suggesting more programs and websites for working with each file type, providing in-page previews of files, etc, etc.

But before I do any of that, I thought I'd just get version 0.1 up and ask for some feedback. Any questions or comments are very welcome. By the way, you need Flash (at least version 10) to make it work.

I tried it with an epub file before I read your comment about its mimicking the unix 'file' command, and was disappointed at the output.

I'm in the process of writing a routine for checking the validity of epub files (i.e., to unzip the archive, make sure the required directory structure and xml is there, etc.), and at first glance, that's what I thought your app would be able to do.

Yup, that's what I meant by improving accuracy. Unix's 'file' is a good start, but it could definitely be improved on.

Well, in addition to epub, it would nice if it could validate other complex file structures, e.g. docx or any of the other OpenXML formats: http://openxmldeveloper.org/default.aspx

Perhaps you could make money through affiliate marketing. Let's say someone asks what a svg file is. You could provide links to svg->pdf utilities.

I gave it a .js~ to see what would happen. And it said: "That file contains ASCII C++ program text"

It should probably say it's a javascript file. Or maybe:"The extension of the file is javascript the content appears to have ASCII C-style syntax...so it probably is indeed javascript." Although worded better. Like someone else said a lot depends on the market you're trying to reach.

Have you considered adding a statement about privacy, ie what you do with the file once it has been uploaded? When/how it is discarded.

That's a good point (the files are immediately deleted btw).

(quoting what you said in another answer):

I'm uploading the first 10KB of the file

Mentioning this upfront would probably help.

I can't see a lot of use for this as a webapp.

If I were going to use this, I'd want it where my unrecognized files are. If you could connect this to a right-click in my filesystem, or have it as a browser extension that intercepts downloaded files and tells me what they are (ie: built into firefox's download manager, or the download bar at the bottom of chrome). Or get it to interact with my email somehow. Next to attachments, have a link saying "what is this?".

Not sure if any of those are actually possible, but that would upgrade your side-project from a "cool idea" to something that I would actually use.

If it's gonna be aimed at techies, it'll have to be geekier, if it's gonna be aimed at regular humans, it's gonna have to be friendlier (ie. "It's a Visio Doc..." vs "It's a microsoft office OLE doc"). I guess that's the main choice to make as to direction right now.

My thoughts exactly. Perhaps it could do both based on extension though, e.g. I doubt you'll find a "techie" not knowing what a .docx is so that info could be something like "This is a Document for Microsoft Word 2007+" Etc, while a rarer or more technical file e.g. .torrent (yes, I know that's not the most technical file extension) would have more specific info.

Yup, and thinking in terms of "customer development", perhaps you can share some hypothesis use-case of how a non-techie would use this.

Meaning, in what situations would the non-techie say, "gosh I need to know exactly what file this is?"

Well the first thing that come to mind, would be an extension they didn't recognize and couldn't open? Like .ppt but they don't have PowerPoint installed. Of course, this somewhat refutes my earlier comment as well, because most of the unrecognized extensions might be more technical in nature.

Perhaps there should be a top sections with something like: "I like it dumbed down for me: .abc is a abc application file" and a section below that for "I'm 1337, give it to me straight: this .abc file is a windows me executable with blah blah blah..."

Aside from choosing better titles for the sections, that might be more helpful to everyone. All of the file detection only needs to be done once.

It needs to have a significant differentiator from just googling the file extension.

What if the file extension is non-existant, incorrect, or uncommon? This site handles that.

For instance, TaskPaper stores its todo items in a file with the extension ".taskpaper" -- googling for that will tell you that the file comes from the TaskPaper app, but it won't tell you that the file is really "ASCII English text"

Maybe that just implies I should aim it at people who don't know what "google the file extension" means? ;)

Maybe you hit the SEO hard enough and you are on the receiving end of "google the file extension".

If I give it an Office 2007 file, it identifies it as a zipped XML file. While this is technically correct, it's probably not the most useful answer. This can probably be corrected with just file extensions, but looking up XML namespaces would be an interesting idea.

Was wondering: Why do you need to use flash for this?

I'm uploading the first 10KB of the file and identifying that on the backend - from experimentation, 10KB is enough to identify most common files. Flash lets you do this easily, whereas I'm not aware of any other web technology that would allow me to do it at all.

Was wondering: Why not use flash for this?

Jeezus, Flash bashing is getting real old.

I am not trying to bash flash here... the site seemed simple enough to me to be implemented without requiring flash, and I was curious as I had to click an extra time because of flashblock.

i'm not able to have it successfully recognize a file. i sent it a txt file and about 2 minutes later (finally) it returned with "Sorry, couldn't identify that file." ?

What version of Flash do you have? Right-click on the button and choose "About Flash".

Maybe something useful could be "Fix this 'corrupted' file".

hehe I like the commentary. very friendly and responsive.

Whats next on the agenda?

Reply on news.ycombinator.com