Re: 2 depend or not 2 depend
On Feb 13, 2010, at 9:30, Nicholas Clark wrote: [Checking binaries into VCS] This system likely *wouldn't* scale if we needed even a second architecture. We have a git submodule for cpan/. We primarily deploy and develop on x86_64 linux, but we also keep the darwin/OS X arch updated so we can run things on our laptops. Graham made http://github.com/abh/combust/raw/master/bin/diff_cpan_arch to easily show which .pm's are out-of-date. ~10 years ago at ValueClick we had Linux and FreeBSD binaries of everything (perl, apache, ...) checked into perforce to accomplish the same thing. Anyway; I agree that it's much nicer than trying to use .rpm/.deb/... dependencies in your application. - ask
Re: No more IP for you
On Jan 20, 2010, at 6:28, Bruce Richardson wrote: IPv6 offers much much larger ranges and much simpler renumbering schemes. The old mistakes are undone: enough ways to make new mistakes. The demand upon a resource tends to expand to match the supply of the resource., to quote the general application of Parkinson's Law. It may seem that IPV6 has a huge range to give out, but that is only going to encourage people to produce solutions where every light switch and light bulb in the world (and eventually every cell-maintaining nanobot in every human body) receives its own IPV6 address. Once the conversion to IPV6 is passed, the new address space will be consumed at a much faster rate than the old one. A standard end-user allocation is, currently, as many IPs as the full IPv4 address space. The plan is to do it that way for the first 1/64th of the address space or something like that and see how that goes. - ask
Re: No more IP for you
On Jan 20, 2010, at 16:44, Dave Hodgkinson wrote: Isn't it a case of wresting some class-A addresses from the like of IBM, ATT and HP and making them use pukka internal addresses for inside the firms? That'd help for a little while; just like whatever market will come up for IPv4 addresses when no more can be allocated from the RIRs will help for a little while. But really, we're running out. Yeah, not next month. Maybe not even when the predications say, but by definition before everyone's ready. - ask
Re: SHA question
On Jan 15, 2010, at 14:19, ian wrote: My understanding[*] is that it computes a checksum for each block of a file and only transmits blocks that have different checksums. And to calculate the checksum on each block of the file, it has to, um, read each block of the file... yes? Doesn't rsync *push* rather than *pull* in which case the files it computes the checksum on are all local. I did not think it worked in the way you mention without rsync daemon running at the remote end doing the checksum for you. But with NFS the remote is local. You need an rsync box running where the storage is to get cheaper checksums. - ask
Re: Domain acquisition
On Dec 14, 2009, at 4:09, Jurgen Pletinckx wrote: But what is the etiquette in these situations? I'd rather not reveal to them to what extent my friend is interested in the domains. To hide that I have to go through aliases or proxies. Which feels just a bit sordid, somehow... 1) Offer more money. No reason to reveal who the real buyer is or what the domain is for. 2) Consider if the recipient is actually getting your message (assuming they're not responding). There was an unused domain we wanted where we offered $1000 (or something like that) but never got a response. Some time later the domain expired and we could get it for $70 or some such from the registrar that controlled the almost-deleted domain. - ask
Re: Perl Christmas Quiz 2009
On Nov 30, 2009, at 13:21, Martin A. Brooks wrote: A sysadmin and pedant's point of view I take it sysadmins are too angry and bitter to understand or care for the holiday[1] spirit I'm sure the quiz was sent in. Why don't you go change someones password? - ask [1] http://xrl.us/holidays
Re: Efficient sorting of SNMP oids
On Oct 31, 2009, at 15:17, Dave Hodgkinson wrote: Using Sort::Key::OID or sort::maker would have been great, except that we need to deploy the script to a rather large number of servers and it is traditionally long and painful to get modules approved and deployed in our machines. Then build it in to your own dist tree. Yeah; that's what we do. We have a git submodule (cpan/) with our application, so when we upgrade or install a module it gets branched/ merged/tracked/code-reviewed/managed just like any other application change we make. We have a helper script[1] that starts the CPAN shell with INSTALL_BASE appropriately configured to make it easy. - ask [1] http://git.develooper.com/?p=combust.git;a=blob;f=bin/cbcpan;h=b69b81829e43450344f9dc35349151cf2750e6b5;hb=HEAD
Re: Credit Cards
On Oct 14, 2009, at 13:39, the hatter wrote: On Wed, 14 Oct 2009, Avleen Vig wrote: Here in the US, when you place a credit card order online, you almost always have to give the phone number associated with the credit card account, and that is definitely verified by the CC company / bank. I see no problem with this really. Of course, being a security question, you wouldn't want to set it to something that anyone with a phone book can look up. While they use it to match; they can also use it to call you on a verified (by the bank) phone number and confirm that you (the account owner or someone in the same household) really did place the order for 47 chainsaws to be shipped to Nigeria. - ask -- http://develooper.com/ - http://askask.com/
Re: Anyone hiring at the moment?
On Sep 29, 2009, at 8:26, Ricardo Signes wrote: ...but there's also a large culture of beer ignorance. Once at the local Beverages and More[1] where they have hundreds of different beers from around the country and the world I overheard a couple talking about choosing between Miller and Budweiser beer. I wanted to scream! - ask [1] http://www.bevmo.com/ - the and more part of the name is really just to make it sound like a store name. Beverages would be more accurate.
Re: Anyone hiring at the moment?
On Sep 21, 2009, at 15:16, Abigail wrote: I've been bumming around contracting for the last few months - but I think that a regular salary probably suits me (and my slackness about invoicing) better. Is anyone interesting hiring at the moment or has this financial meltdown thing (news of which has reached even the provinces) stopped all the fun? Well, we (booking.com) are. But that unfortunally requires a relocation to (or near) Amsterdam, and I can't imagine anyone willing to do that. At solfo.com we hire skilled Perl people once in a while too, but we're even farther away! (Los Angeles; although only half of us are there) :-) There are a good number of Perl shops around here actually; I think it's some combination of a reasonable number of big shops using Perl and enough people leaving those places to start new companies (and then using Perl). In other parts of the country not so much; you need a certain pool of jobs and workers to keep it going (and growing, hopefully). It's probably the same in London vs outside-London. The best solution? Start you own company and hire Perl developers! :-) (I'll leave the missing steps as an exercise to the reader). - ask
Re: Effort-free, box-it-up paperwork management
On Aug 23, 2009, at 16:54, Dave Hodgkinson wrote: Prove you haven't dicked with a digital copy. (Knowing this is London.pm): In the US the taxman actually specifically allows reasonable digital copies. It's sorta vague what a proper copy is; I'm guessing so if it's a terrible scan that looks the least bit suspicious they can just say they won't allow it. (Which in turn is why my accountant here also tells me to keep the paper versions in a box somewhere). This is the future; we were supposed to be done with paper already. Also, jetpack! - ask -- http://develooper.com/ - http://askask.com/
Re: Decent OS X audio rip software
On Jul 27, 2009, at 11:19, Paul Makepeace wrote: I've just discovered that iTunes will merrily rip CDs with errors and make no mention of this. Is there a way to have iTunes bail or retry on rip error? I've no idea what it does, but there's a Use error-correction on Audio CDs checkbox in the preferences. I haven't tried it, but I looked at Rip and Max for my never- happening re-encode-everything project: http://sbooth.org/Rip/ and http://sbooth.org/Max/ - ask -- http://develooper.com/ - http://askask.com/
Re: [OT] Wrapping methods
On Jul 6, 2009, at 9:06, Nicholas Clark wrote:
Of course this can also be solved in various other ways, such as
having the
code be
@ISA = 'Generic::Base::ClassWithTimeout';
sub run_with_timeout {
... # does stuff
}
and Generic::Base::ClassWithTimeout having a run() method that calls
$self-run_with_timeout(...) inside the alarmed block, but that
doesn't feel
as elegant an interface.
I'm pretty sure Moose can help with this sort of thing.
I think you can get the API reasonably elegant with just clever naming
though. Don't make it run_with_timeout, but _run (or whatever -
something generic). The Generic::Base::run can call that
automatically so it'll Just Work for all the classes (with or without
wrappers).
- ask
Re: Java wonks?
jobs.java.org? Weird ... that doesn't work. :-) - ask
Re: Hardware Reliability
On Jun 5, 2009, at 19:34, Avleen Vig wrote: All is not what it seems. Take this with a spoon of salt. That is so unfair; you must tell more. :-)
Re: Hardware Reliability
On Jun 4, 2009, at 13:49, duncan.garl...@ntlworld.com wrote: Can somebody please point me in the direction of some authorative reliability statistics for server hardware, preferably including add- ons such as disc arrays? Google has a paper on that stuff for individual disks. In my experience failure rates for everything else is much lower (but each failure of course is more hassle): http://labs.google.com/papers/disk_failures.pdf - ask -- http://develooper.com/ - http://askask.com/
Re: Best practice for releasing Perl modules to staging and live
On May 19, 2009, at 7:22, Paul Johnson wrote: The way I do this is to build perl and all the modules I need on the development machine. Then I package up the whole lot as a single package. The big failure in this is that you can only have one version of stuff. One developer experimentally wants to try a new version of Foo::Bar? Too bad. Want to deploy two versions of your app on the same box (prod/stage running as different users). Too bad. One developer debugging an older version of the application that a partner is using? Too bad. Perl modules are part of your application; not of the system. - ask
Re: Best practice for releasing Perl modules to staging and live
On May 18, 2009, at 6:22, ian wrote:
How best to ensure different environments have the same versions of
Perl modules?
A CPAN module is part of the application code, so treat it as such.
We have a Perl installation with a basic clean perl install; no
extra modules. Our application has a cpan submodule (yay git) where
we install all the CPAN modules we need. This way all branches (in
development) and releases (for staging and production) include exactly
the module versions we need.
We have a couple helper scripts to make it easy:
CPAN shell that sets the paths just right ('CBROOTLOCAL' is the local
application path):
http://github.com/abh/combust/blob/master/bin/cbcpan
Wrapper that starts the right version of perl (configured in a
configuration file) with PERL5LIB set correctly:
http://github.com/abh/combust/blame/master/bin/perl
Check that we've installed the same versions of a module for all
architectures (we deploy on Linux, but do some development on our macs):
http://github.com/abh/combust/blame/master/bin/diff_cpan_arch
- ask
--
http://develooper.com/ - http://askask.com/
