Disqus - Latest Comments for vitaly

Re: Astrails How to fix a hosed /etc/sudoers file on Mac OSX

vitaly — Sun, 29 Dec 2013 04:26:53 -0000

The problem is not with permissions, 0440 are the right ones for /etc/sudoers file. The problem is with broken syntax inside file, and since this is the file that can grant you extra permissions with 'sudo' it was a problem to fix.

Re: Writing Markdown With Style in Vim | Astrails - Hi-end web technology

vitaly — Mon, 14 Oct 2013 08:53:07 -0000

WAT? :)

Re: Writing Markdown With Style in Vim | Astrails - Hi-end web technology

vitaly — Mon, 14 Oct 2013 08:51:14 -0000

well, I usually do **not** want all my markdown files look this way ;). for example, not when I'm just editing some api docs in my project. I do want this when I'm editing contracts, or blog posts - when I'm working on a single file.

Re: Writing Markdown With Style in Vim | Astrails - Hi-end web technology

vitaly — Mon, 14 Oct 2013 08:50:16 -0000

well, I thought about it but decided not to waste time. when I'm using it Im usually in a 'focus' mode anyway. i.e. working on markdown document, like a blog post or contract. usually no need to mix it with source code buffers. so using a separate window makes sense

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Thu, 25 Jul 2013 06:56:28 -0000

I do know about the append only with sync, but its much more then that. any sane payment processing system on postgres would use transactions and locks, and I can't think of a non-completely-stupid way to implement payments with postgres so that it would keep charging customer's card over and over. you would fail much earlier if something went wrong. here it a) lost the balances and b) went into read-only so that it kept charging and charging and charging. something that wouldn't happen with postgres as easily, as for one it wouldn't get into this read-only state.

in general RDBMS systems were developed and tuned for lots of years, Redis has a loooooong way until it will be as 'safe'. I'm very much for it succeeding, but Im as sure that it still has problems that just need some time to be found and fixed (and a couple more incidents like this along the way).

for example, the whole "restarted with wrong configuration" was based on a feature that was implemented with good intentions but used without enough thought. I know its improved now, but it doesn't take away the fact that it was very very easy to get into this situation, where your config files on disk differ from the configuration you are actually running with. simple restart and boooom.

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Thu, 25 Jul 2013 02:32:40 -0000

and reading the description of the events I can't imagine the same kind of payment process and failure scenario happening with, say, Postgres. Not unless you do something incredibly stupid. And stupid those guys are not. The db they used let them create such a system and use it in this way.

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Thu, 25 Jul 2013 02:27:40 -0000

All I'm saying is that data consistency and persistence guarantees of Redis are not good enough to deal with actual credit card transactions. I think an RDMBS is much better suited for that. I actually love the new kind dbs, I'm especially looking at Redis and Riak, but I know that for payments I'll still definitely go with postgres. even if most of the application is in a nosql data store.

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Thu, 25 Jul 2013 02:23:39 -0000

I definitely seen, but the failure scenarios are different, and much more then 'we restarted the server' is needed to lead to a data loss.

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Wed, 24 Jul 2013 13:03:05 -0000

No its not. It might be ok to hold the 'amount the customer owes us' number, but the "current customer's balance" and actual payment transactions should not be held there. In such case the worst that can happen is that you will under-charge your customer in case you lost this data, and not over-double-tripple-charge-until-credit-card-is-locked that happened in this case.

Re: Billing Incident Post-Mortem: Breakdown, Analysis and Root Cause

vitaly — Wed, 24 Jul 2013 05:00:23 -0000

I've always said that billing should stay in RDBMS. NOSQL is a bad choice for balances and payment transaction data.

Re: Patching Ruby for Faster Rails Startup, Revised | Astrails - Hi-end web technology

vitaly — Thu, 16 May 2013 15:48:19 -0000

While 2.0 is already there, we still use 1.9.3 on all the projects. That being said, I'm pretty sure we'll switch sooner then later. Expect an update blog post then ;)

Re: rvm install patched ruby for faster rails startup | Astrails - Hi-end web technology

vitaly — Thu, 15 Nov 2012 02:50:52 -0000

did you do the `rvm get head` part before the installation?

Re: Mac OSX Tools: LaunchBar from Objective Development. | Astrails - Hi-end web technology

vitaly — Thu, 16 Aug 2012 17:54:25 -0000

UPDATE: I did try Alfred now and I actually like it a lot. Trying to use it for a while instead of LaunchBar.

Re: problem with running spec in vim - Avi Tzurel

vitaly — Mon, 06 Aug 2012 13:33:08 -0000

how did you figure out its vim-space and do you know why exactly does it break it or how it is related to the problem?

Re: Mac OSX Tools: LaunchBar from Objective Development. | Astrails - Hi-end web technology

vitaly — Wed, 18 Jul 2012 02:57:57 -0000

no, didn't try it, I settled on LaunchBar, but you are the 2nd person mentioning Alfred to me today ;)

Re: http://railsdog.com/blog/2010/10/redirect-non-www-requests-the-rails3-way/

vitaly — Sat, 16 Jun 2012 18:44:43 -0000

very nice. thanks.

Re: Windows 8 may drive me to Linux

vitaly — Mon, 05 Mar 2012 19:46:28 -0000

do yourself a favor and switch to OS X instead of Ubuntu. Its still Unix underneethe and you can DIY what you want, but at least it "just works" and you don't have to fiddle and configure it on an almost daily basis. Linux is for servers, OS X is for desktops, Windows is for corporate accounts that can't upgrade ;)

Re: CoffeeScript: Spartan JavaScript

vitaly — Mon, 22 Aug 2011 15:08:47 -0000

I think yes, on, no, off are not for use in the case like `item.has("title") == yes`

I think its more to be used with another alias keywords like is and isnt, e.g. `if verbose is on` or `if debug is off` etc.

Re: Astrails How to fix a hosed /etc/sudoers file on Mac OSX

vitaly — Mon, 13 Sep 2010 23:50:37 -0000

there is no getfacl on my snow leopard install.

may be its a leopard thing.

Re: A Designer’s Comfort Zone

vitaly — Mon, 30 Aug 2010 18:20:32 -0000

Well, as you said, you are more satisfied with the results when you do stay within the comfort zone. So as a client I'd prefer you not to experiment my behalf ;), because the result is not guaranteed to be the best you can do. Just create the best thing you know to do *right now* and learn new things in your free time ;)

Re: Cassandra and HBase Compared

vitaly — Wed, 21 Jul 2010 17:26:52 -0000

> furthermore adding a keyspace requires a cluster restart!
This is simply not true. Yes, each node in the cluster needs to be restarted, but this doesn't mean AT ALL that they all should be restarted AT ONCE! you can restart them one by one without loosing a bit

Re: Are We Too Complacent About Cloud Computing? [VIDEO]

vitaly — Tue, 22 Jun 2010 08:22:18 -0000

update: more importantly the term means whatever the people that are using it mean.
Its been a looooong time since Linux meant just the kernel.
For many years Linux means both a kenel AND an collective name for a complete OS composed of "Linux Kernel", GNU user-land, a ton of Apache stuff and a gazzilion of independent opensource soft.
Actually its even more then that, I'd say just "Linux" no longer refers to the kernel. If you want to refer to the kenel specifically, without the userland and stuff you actually better call it "Linux kernel" or people might be confused (depends on the context though, if you compare Linux to Hurd or Mach it'd be ok for example).
The lunatics leading the GNU/Linux fight should come to the terms with reality. They have lost. Just a couple of distributions *chose by themselves* to call it GNU/Linux, the rest of the world is pretty happy to call it just Linux. RedHat is RedHat Linux, not RedHat GNU/Linux. Same for Gentoo etc. Hell, even Ubuntu is NOT GNU/Linux. From the major used ones Debian is the only one. Get over it already, stop wasting time of everybody on the lost fight which wasn't important to begin with (well, apart from RMS' ego and unreasonable demands for everyone to promote GNU that is).

Re: Are We Too Complacent About Cloud Computing? [VIDEO]

vitaly — Tue, 22 Jun 2010 08:08:21 -0000

This is just another example of the same ignorance. The OS is called whatever the OS creator call it. If I want to call it Vitalix I can call it that, and no one, especially not RMS has any right to spam my mailing lists to demand I call it GNU/Vitalix ;).
This is exactly what I remember from the days of being on the Gentoo Linux devs. Its Called "Gentoo Linux", period. There's no GNU in the name, and will not be. But as usual RMS came and started his annoying persuasion campaign.
Debian guys can call their OS however they wish.

Re: Are We Too Complacent About Cloud Computing? [VIDEO]

vitaly — Wed, 09 Jun 2010 15:26:09 -0000

he is basically a bigot. and like any other radical zealot he's hurting his own cause. He makes Open Source look bad. I still remember the times when he was practically spamming every little linux distribution mailing list with demands to call it xxx gnu linix. wtf? why not xxx gnu/apache/and_million_hackers linux? :)

anyway, his obsession with 'control' is just sick. guess what? I don't have time nor do I want to manage my own email server. I did once. I've run qmail and postfix on my own linux servers, but now I just outsource it all to gmail. it works. it saves my time. suck it up.

and what's the point of using his pathetic 9" netbook? What does he prove except for his own craziness? Wouldn't the open source world benefit if he'd just get a decent hi-end laptop and hack on some software instead?

Re: Astrails SmallRecord - Simple Object persistency library for Cassandra

vitaly — Fri, 30 Apr 2010 18:40:54 -0000

this is some point-in-time freeze of the activemodel. I'm actually planning to remove it from vendor and add a gem dependency once rails 3 is out.