Redis on Windows prototype | Hacker News Reader

Redis on Windows prototype

github.com

47 pointsyread24 comments

I'd be much more impressed if they just spent some time on their POSIX layer (which exists and is close to tolerable), so that arbitrary projects would compile. Why massively hack one project when you can make multiple work?

For a fairly large number of reasons, actually. Lets say you wanted to start the work now. The POSIX layer is probably owned by the OS division. They're pretty busy finishing up Windows 8. New versions of this layer would almost certainly require a new version of Windows, and as it would be too late to get into Windows 8, the earliest it would ship would be in the following release of Windows. Additionally, the OS team probably doesn't have much of an incentive to improve the POSIX layer, whereas whichever organization is funding this project is likely to have goals which align much more closely with the work being done.

People sometimes seem to think that Steve Ballmer is Steve Jobs: micromanaging these kinds of decisions from the top. Instead, they're often made fairly far down in the org chart

You're completely right.

I also feel bad. I accidentally did one of the things I hate, off-topic comment on actually cool project, and would delete my post now if I could.

I commend everyone who worked on this for pulling it off - it can't have been easy. And if it allows them to offer redis on Azure, that's awesome, as long as it keeps current with redis proper.

It doesn't matter at all for consumers really, but in the server space I think the inability to run great open source applications is a bigger issue than even licensing cost. I hope, likely in vain, that they'll address this at some point in the future.

The comments here seem brutally negative, but I applaud Microsoft for yet another step in what seems to me to be a firm-wide aggressive embrace of open source tech. The Microsoft of yore would not have built this at all, let alone released this on Github.

> The comments here seem brutally negative,(...) firm-wide aggressive embrace of open source tech.

Some of us still remember how Microsoft used to Extend and Extinguish right after Embracing. Protocols, formats, standards; you name it, they did it. It is true the company seems changing to better, but you can't re-build damaged reputation overnight.

For example, http://news.ycombinator.com/item?id=3900348

> The comments here seem brutally negative (...)

Those aren't that brutal yet. Read up some discussion on software patents. Or on PHP ;-)

This is inline with their work to get hadoop, Linux etc. They are not adding any new featues. They are simply porting to their proprietary operating system.

Supporting a major new operating system is not a new feature for any software?

see also http://blogs.msdn.com/b/interoperability/archive/2012/04/26/... for more details and reference to dmajkic project. The main difference here seems to be they managed to do the asynchronous persistence without a fork. They also mention in the comments that they are planning to release it for Azure.

Previous HN discussion: https://news.ycombinator.com/item?id=3333298

I was using this, https://github.com/dmajkic/redis, previously.

can someone post a pointer to the diffs?

Here's something to get started:

https://github.com/MSOpenTech/redis/compare/antirez:2.4.11.....

@antirez said: sorry, NO THANKS!!

No I did not ;) What I think is:

1) It's cool to see Redis ported on win32, especially now that they are trying to provide an implementation that can persist asynchronously and so forth.

2) Not interested in backporting changes into the main Redis codebase at the moment, because there are little reasons to deploy on WIN32 IMHO, and the changes to the code base are massive.

So it's interesting, it's listed in the "download" page of hte official site as an external project, but there are no solid reasons to make the official code base more complex.

now i see! thank you for your great works :)

because there are little reasons to deploy on WIN32 IMHO

There are plenty of reasons to deploy on Windows, and it has lost Redis a lot of momentum and wins that it doesn't support it.

Generally speaking there are two things that can happen if Redis win32 support is put in the main code line: it succeeds on win32 or it fails on win32.

If it succeeds on win32 then, based on previous history, we can assume that similar functionality will be added in some oblique fashion to Microsoft SQL Server. At that point Microsoft would, based on previous history, withdraw support from Redis so it could provide its customers a better experience by concentrating all of its resources on the MS SQL product. Lacking support, Redis on win32, would whither and die on that platform.

If Redis on win32 doesn't succeed then one would expect Microsoft to withdraw support and Redis on win32, lacking support, would whither and die on that platform.

Edit to add Not saying Redis can't succeed on win32, just that it can't be completely dependent on Microsoft to do so.

We've had this discussion a while ago. The outcome is that it's antirez's call. There's plenty of us who'd like him to take out a few months to drop new features and do windows portability instead, but it's his call. If antirez would always take the easy way after some user space resistance, he'd be a mysql contributor.

The outcome is that it's antirez's call.

Absolutely true. He owes me or anyone else nothing. I am not saying that he should do it, I'm just replying to the specific statement about deploying on Windows. There are plenty of reasons someone might prefer to deploy on Windows, which explains why almost all successful technologies have first-class windows options (Mongodb, PHP, Ruby, Apache, MySQL, postresql, Membase, and on and on and on).

I'd love to see some figures or even anecdotes backing that up.

That's not snark- I'm genuinely curious.

There are 88 million (~15% of total) web sites hosted on Windows (http://en.wikipedia.org/wiki/Web_server#Market_share). That's 88 million reasons.

If you're running your web server on the same host as your redis server, in my experience you're doing it wrong.

That doesn't necessarily mean that they run it on the same host. It simply means that for some reason a lot of people have chosen Windows as their web platform (which could be as simple as familiarity). It is foolhardy to dismiss that massive platform.

Next week from the Cool Stuff ported to Lame Platforms department: Node.js on OS/2

Reply on news.ycombinator.com