Hacker News Reader

Comment by _pi | Hacker News Reader

I'm surprised more companies don't use version control such as git in their documents.

Most companies use Word for their documents, which doesn't lend itself too well to the standard version control scheme since it's a binary file.

Then there's the whole open source thing that certain companies are afraid of (not worth going into or arguing about, but it's there). Some places do use Sharepoint, which will do versioning on Office files, even though it sucks, but it does cost a lot more!

Ultimately though I prefer writing documents in a wiki, though I might look at something like Github for that now too.

OOXML is text based now. So Office 2007 should be text capable. Sharepoint is generally horrible with that I agree, but I think it uses the default MS word mechanism to do that which stores changes with the file at the end.... kind-of a bad idea.

Also fun fact: Sharepoint agreeably sucks more than any other MS product but it's not actually written by MS, but a large part of it is written/owned/supported by Akamai Technologies

OOXML is typically stored as a compressed ZIP archive. Even when its decompressed to its constituent XML files, modifications are performed without much regard to how a diff will look.

This applies to auto-generated XML in general -- ODT is not immune. LaTeX and reStructuredText are the only formats I know of which are suitable for professional publication and can be usefully diffed.

That's very interesting. Why is auto-generated XML hard to diff?

Consider an XML file consisting of one long line of text. A diff would not be very useful in that case.

I guess an XML aware diff could work though? certainly not at all line based...

If it did node-based diffs.

Most diffs do that. The Diff that ships with gvim on Windows has this option on by default and I couldn't find where to turn it off because it was severely screwing up what I was diffing.

That's because XML does not guarantee order of elements.

"That's because XML does not guarantee order of elements."

It better, or else it's broken. In XML, element order matters. Attribute order does not.

I'd be interested to hear where you get the Akamai Technologies-SharePoint link, I've only heard that SharePoint is wholly written by Microsoft.

Very weird support call. I finally said listen I think it's a problem with Sharepoint, and they forwarded me to Akamai, however they answered the phone as MS, but the CallerID showed Akamai Technologies.

If you've ever dealed with MS Business Support they have sections for every single product and some products are segmented by parts.

Oh. MS's SharePoint premier support is definitely staffed with contractors; the two times I called, I talked to non-employees. I wouldn't draw any conclusions about the development staff from that though.

Akamai's services are almost too tightly integrate-able with sharepoint. Not only that I wouldn't call the support premier ;-)

Our company, myself included, LOVES using Apple's Pages for documents. Unfortunately, .pages documents are actually directories full of files, so you can't check check in a file and then view it/download it using a web repository browser.

Upgrade to iWork '09, which uses single files.

Very nice! I'm actually using '09 but hadn't noticed that they had changed that. Awesome. Thanks!

I think some employees either can't be bothered or don't understand it.

I do writing as well as coding for personal stuff, and I know I love putting my text into mercurial to track changes. I figured I couldn't be an outlier but maybe I am!

As do I. I actually am building a two tiered system as my server parts come in. I pretty much have a .repo file that says to track which folders with what protocol either git or rsync, that way I won't fuss with stuff that;s annoying or stupid to version control like binaries, and music.

What's a pain is trying to get those protocols to work not over FTP because SSH slows down file transfer by a factor of 10. When I do transfers via SFTP or SCP I get about 1.1-1.8 MB/s when I do it via FTP I get about 10-11.8 MB/s.

I don't think that is a problem with transfers over SSH. The compression wouldn't cause it to slow down speeds by a factor of 10 so there must be something else that is cutting down the speed.

I regularly (and only) transfer files over ssh/sftp and can manage to sature my 20mbit cable connection.

Hmm, that's weird, I get much higher speeds in vsftp what protocol and settings are you using?

Take a look at the mr command. I find it very handy for working with heterogeneous collections.

http://kitenet.net/~joey/code/mr/

Same. I use subversion to keep track of .tex and .svg files as well as many other odds and ends.

I was regarded as some strange kind of mutant because I not only kept academic papers in version control, I also had a single global BibTeX database that all my papers referred to. It was really, really nice but so not the done thing in the department I was at.

It is pretty complex, people are used to doing it manually: 1st draft, 2nd draft, 3rd draft, etc.

It might be a good product idea, an ms word clone that brings the benefits of version control.

Also, most common rich text formats don't play well with the diff format, so it can be difficult to see changes in a meaningful way. I actually hope to work on a solution to this problem as a school project this fall, so if anyone has any suggestions, I'd love to hear them.

Other than that, most version control interfaces leave something to be desired. They are natural for us, but not for most other people. I realize that there are some good tools available, but I haven't seen enough of them to say whether or not average users could get used to them.

For my "thesis" like paper I had to write for university to finish my degree I learned LaTeX just to be able to keep my document under version control.

I didn't actually use very much of the vcs, but the added security gave me enough peace of mind to just not worry about everything.

Plus, not having to hit apple+s and upload/mail/transfer it manually to another computer everytime, to "backup", saves a lot of time as well.

The new MS Office format is XML with plain text. RTF is also plain text.

However it wouldn't be very hard to do, if you implement a small script that takes the git history and pushes it through a viewer showing different versions of the file. Obviously highlighting would be harder but the changes would be visible. Highlighting requires a bit more babysitting with the file format to produce valid files obviously.

That's the issue I was referring to. The diff tool works on lines, but changes to formatted text documents are word- and tag-based.

I'd say that viewing the changes are more-so the problem because diff will be able to handle the backend of things on most files, changed words in lines are picked up easily, so are tags.

Word's change tracking gives people history. No merging, but most seem able to live with that. Worst case some admin assistant has to piece all the changed documents together by hand.

A lot of companies use wikis, which track modifications.

Reply on news.ycombinator.com