Bob Lee › Comments

I talked to a guy at PARC at a Google Dev Day who said they have a system like that, even said it was open (or was going) source. I forgot the name of it though. - Ray Cromwell

I've always found it shocking that no one's built anything like this. It wouldn't be all that complex, and would be incredibly useful. If you think of the name of that PARC project, let me know! - Joel Webber

It would be nice if I could just take a pic w/ my phone and it would upload to my database automatically. - Bob Lee

You've been doing Android stuff lately, haven't you Bob? :) - Joel Webber

Evernote would let you take a photo and it does OCR against the picture. Works pretty well. I use it for whiteboards. - JR Boyens

A Fujitsu ScanSnap does a good job of handling skewed and weird documents. I've got one at home and I digitize a good deal of our stuff that way straight into DropBox. Does OCR and creates searchable PDFs. - JR Boyens

@JR: If it creates PDFs, I guess it's not fuzzy-match searchable, no? Sounds like a reasonable interim solution, but I think it would be wonderful to have it do best effort matching on searches. I mean, I'm sure the OCR algorithm generates a probability distribution for each letter/word/whatever, which would be hugely useful for this purpose. - Joel Webber

Well, the PDFs are OCR'd using FineReader or Acrobat's OCR method. I've found that Acrobat is MUCH better than FineReader. If you're asking about partial string matching, it'll do that. If you're asking for wildcards / regexps, it could do that I suppose so long as you could find something that would read the PDF index. - JR Boyens

I think he wants OCR-aware indexing instead of orthogonal text indexing. Converting to text and then indexing means I can only search for that exact text result (which may be an incorrect guess); it would be nice to search for images that may match the text instead. - Bob Lee

It is unbelievably bad as far as that is concerned. Great back end with a shitty user experience. I don't think it has to be that way. - Sam Pullara

One big problem I have is unclear, inconsistent and overloaded terminology -- words like "branch", "change", "commit", "index", "add", "checkout" and "label" all mean different things in different places. Even the official documentation tends to mix and match words. This is OK if you know how everything works, but when I'm trying to get your head around the dozens of places and ways git keeps state, it just leaves me confused and unhappy. - ⓞnor

The good thing is that you only need a hand full of commands most of the time. I use a cheatsheet for the more esoteric ones. http://gitready.com/ has some good articles explaining the different states and the articles are organized by difficulty. - sdfx

Should I use mercurial instead? - Kevin

Kevin, I think Mercurial is easier to use - Paul Franz

Things I don't like to hear about tools I use: "the articles are organized by difficulty" - ⓞnor

Agreed, but to be fair, Linus said at one point that he hadn't originally intended it to be used directly, but rather for other tools to be built on top of it. As is often the case among *nix geeks, people just started using the low-level tool because I guess no one felt like building a user-friendly version. - Joel Webber

Linus was kidding himself, I don't think it's possible to make a user-friendly wrapper. That would be like making a user friendly tool on top of Unix shell utilities -- people have tried it, but it just doesn't make sense. The underlying commands are just too generic and powerful. So what we end up with is a bunch of "leaky abstractions" -- helper layers and "porcelain" commands that try to simplify your life but just add more complexity in the end. - ⓞnor

I've heard C++ STL was the same way — that people were expected to build friendly containers on top of it, but they ended up using it directly, and it's sort of a pain to use directly. - Amit Patel

I don't find git to be too unfriendly now that I understand the basic concepts underlying git -- it is all built upon blobs, trees, and commits. I agree with @nor, sometimes the documentation isn't consistent enough, but I think the situation is improving. - Travis B. Hartwell

I have used higher level tools and I use git on a day-to-day basis. In this case, I made a mistake (probably used "--amend" when I shouldn't have) and wasted an hour paired with someone else trying to undo the damage. We ended up reseting, starting over and retrieving my change from IntelliJ's local history. - Bob Lee