561 karma · joined February 20, 2007
But obviously, they did some tradeoff analysis and decided to do what they did.
Now, from a technical standpoint, the fact that they implemented this feature in the present way results in some information of their internal infrastructure and engineering organization to leak through.
The 5 second, non-configurable timeout should allow us to speculate a little bit on what Gmail's architecture is.
The only reason I can think of is that maybe the undo operation results in a scan over all queued email for all users. This is obviously an expensive operation, so maybe they cannot easily increase this delay with their current architecture. They'll have to add per-user queues on their servers. Maybe that's too big an engineering project at present.
What do you think?
Microsoft Outlook has had a sophisticated version of this for a long time. I know a guy who had set up Outlook to send all his email 30 min after he hit send. You can even schedule a message for sending at an arbitrary time (i.e., per-message level granularity of send delay).
For instance, the Microsoft Outlook implementation of this feature essentially implies that each client can queue up emails for later delivery on the server and later interrogate this queue in a sophisticated way (i.e., it's not fire and forget).
In particular, I like that I can just type whatever I want as a status message instead of Facebook deciding that my message should be 'XYZ is <fill in>".
And I use Facebook instead of Twitter only because most of my network uses FB and not twitter.
I'm unsure if Twitter itself will make it big, but twitter-style updates are going to be big.
If you don't want to be marginalized, fight to get the VPE position. If you're going to get fired or further marginalized for fighting, you might as well learn now so that you'll win next time. If you get VPE, learn the skills required to be a great VPE and prove him wrong. Ask point blank for the position and find out why the CEO thinks you're unsuitable for VPE.
For some reason suits think management is not learnable/teachable. You sound smart; it's quite possible you'll be able to learn the required skills.
IMHO, people who think management is about "people skills" and "politics" are the clueless ones. Read up "High Output Management" by Andy Grove, a hard-core geek if ever there was one. He turned out to be an amazing manager too. Bill Gates, even harder-core nerd also turned out be an amazing manager.
To my mind, management is more similar to programming than to party planning (where people skills are really required). The manager's job is to set up systems and processes which will result in maximum output. To do this, you need to be analytical, quantitative, data-driven, open to being proven wrong etc. Do you see how you've been practicing all of these skills as a techie? Do the same for a proper manager's job skills and you'll see the overlap.
As an aside, I think all managers should have some individual responsibility also. Otherwise they will be tempted to create problems that can be 'solved' to justify their existence.
Re: "politics is not your strong suit" comments: you're expect things to work rationally but they seem to be irrational. Being techie who's always open to the possibility that you might be wrong, you're getting a second opinion to check your conclusions. Nothing wrong with that; that's a strength.
But I have some questions for the the experts. Is git really a fantastic achievement, or is the praise for git coming from people who are not experts in source control (+)?
Based on my cursory readings, it appears that the concepts Git uses (dag, everything is a node, tree snapshotting etc) existed in Bitkeeper.
(+): I ask this because I've repeatedly found that what looks like a big conceptual leap to outsiders is actually a gradual evolution. Examples: I first thought RISC was a big leap, but on reading up the technical papers later, I realized that it was made of lots of innovations by lots of people.
"break them into atomic tasks": I didn't think this sort of hyper-specialization would work in a fast-paced environment. It seems to treat developers as just English-to-C++ translators.
My experience has been that specifying in great detail HOW to do things (as SCRUM does for s/w development) doesn't work as well as specifying WHAT to do and letting people figure out how to do it, but holding them accountable for results. The latter style will naturally result in the evolution of automated systems such as regressions, quantitative user feedback, beta progam etc to ensure that the work got done and quality is maintained.
One should have some structure and avoid reinventing the wheel, but I'm skeptical of detailed recipes like this.
The power of Erlang seems to be due to the fact that processes and crashes are part of the language, and the language makes it fast and simple to create processes and handle crashes. This (esp the process part) can be implemented in a non-functional language.
It wasn't obvious to me why the functional nature of Erlang gave it any power.
It seems to me that things like the Erlang book example where they provide reliability to any arbitrary app can be implemented in C++ using a base class + virtual methods.
I would appreciate any "killer articles" which will make me "get it".
Did the "management gurus" compare that alleged accuracy to FedEx's and USPS's? A person who you know takes a low-value article and coordinates with approx 10 others to deliver it within a few miles by sheparding it on a commuter train. Wow, brilliant! Compare this to Fedex and USPS where hundreds of random strangers cooperate to safely deliver very high value goods through bureaucratic hurdles like customs across countries reliably. Or the systems the USPS has to prevent theft of high-value articles like DVDs, million+ of which are mailed every day.