anybody who posted a new question today got drowned out and away by a wave of edits just to capitalize the words in the title, or just to add a question mark here and full stop there
@NickChammas I think Rolando is just doing major edits and tag edits rather than minor ones now which is what I suggested until we have thrashed it all out on meta
@SimonRigharts ah but are the actual requirements vague :-)
The actual requirements are relatively simple and not at all vague, it just took me three iterations of "no that's not what I want" to get from "I want to see people with high leave balance" to "I want to see the number of people with >160 hours leave balance"
Because I know some of you in here would like to know and stay on top of things:
L5-S1 Intervertebral disc shows degeneration and there is a moderate sized left paracentral disc herniation that displaces the left S1 nerve root posteriorly. The left side of the canal and origin of the left exit foramen are narrowed as well. The right exit foramen is patent. Bones are normal.
L4-L5: There is intervertebral disc degeneration with some disc signal and height loss, but the disc is intact. Minimal diffuse annular bulging is seen. Bones are normal. Canal and foramen are patent.
The rest of my spine appears to be normal.
@NickChammas this has been addressed in an appropriate conversation.
If anyone feels that this has not been addressed, I would ask that further comments be made on an appropriate meta post. Thus far keeping it in chat has been fine, but if further discussion is needed, then I'll have to ask that the community discuss it in a more "visible" position than chat.
@SimonRigharts No, BAs document the business and processes. Actual useful IT analysis is dones by code monkeys. Then when it's wrong, guess who gets the blame
My cynical view is that BAs and the PMs are failed developers and stick up for each other...
@SimonRigharts which is the clbuttic mistake and why explicit JOIN is good
@gbn I've never yet met a BA that disproved the backronym of Bullshit Artist (aside from the really bad ones that can't even do that)
@gbn I'm fully converted to the Church of the Explicit Join, but it's going to be a slow process to spread the One True Path ;)
(I don't actually preach it, but if anyone asks how they should write SQL [not that a developer would ever deign to ask a DBA how to write SQL], I tell them to use the new form)
@SimonRigharts Ah, yes - the myth of the "business" analyst.
The amount of time I spend runing around after BAs and project managers with a pooper scooper represents probably millions of pounds of wasted effort.
Rule #1 of requirements: If they're not ready to turn over to a development team without essentially requiring the developer to re-do the analysis, they're not done.
"High level requirements" is a meaningless catch-phrase invented by lazy project managers who can't be bothered doing analysis work properly.
Maybe, they're not totally useless, as you can get a feel for project scope, but you won't ever get a meaningful project schedule from "high level" anything, and you won't ge able to engage a development team without more or less completely reworking the analysis </rant>
@SimonRigharts Not strictly true - I've worked with good project managers in the past, but maybe 5-10 bad ones for every competent one.
One thing I've seen of good project managers. Without fail, the good ones all come from a technical background and understand the work they're managing.
@SimonRigharts Actually, in integration projects with lots of dependencies on disinterested third parties, I tend to find the concepts of 'schedule' and 'budget' are a bit meaningless ;)
On a straight application development project with no significant integration and well understood requirements, you can make a project plan that at least looks roughly like reality. Integration projects are all about managing external dependencies in my experience. Any project planning methodology that doesn't focus on external dependencies tends to be missing the point.
You tend to find that the amount of actual code that gets written on this type of project is quite small given the amount of analysis and stakeholder management work that actually goes on in the project.
With a bit of handwaving, I use a rule of thumb of about 10K lines of T-SQL code per front office system in an insurance company. However, the project is about 75% analysis and troubleshooting.
@Gonsalu Best viewed as thought provoking fiction.
The best way to treat estimates is to establish deadlines for the third parties, put them on a plan and then start reporting the schedule as slipping when they inevitably fail to live up to their promises.
So, you treat the project plan as a rhetorical prop. Give yourself lots of time, but start slipping the schedule the moment one of your dependencies lets you down.
@Gonsalu No, you have to always treat the data as GIGO. The data owners are the only ones who can clean it up. Report the data, along with any exception reports. You can position it as "empowering them to control and fix their own data", but you should never put sticking plaster over it
With one exception: sometimes you have to use heuristics to match things together where the match isn't recorded at source. In this case you agree the heuristic and specify audit reports for it.
I deal with data quality issues all the time. They're a feature of data warehouse projects. You just report what you've got, and give the business the tools to see the issues and fix them at source. Then they either fix them or don't.
@Gonsalu If the data violates integrity constraints that you need to assume for the ETL process then you can scrub the dud data. I basically find it's worth building a prototype right up front to identify the issues and start dealing with them as early as possible.
Where possible, I build a prototype up front and do a reconciliation piece within the prototyping phase. You can then position it as analysis rather than a UAT bunfight about data quality.
Get is underway as early as possible. If you do it in a quick prototype, you can show the business some early results from the reconciliation - show the data issues and prove you can produce demonstrably correct numbers, or at least figures that reconcile to source.
Then you get them to sign off the reconciliations. It's fairly quick to do this - you can do it with a fairly agile prototyping process.
The reconciliations can then get used as QA tools as well.
The prototype doesn't have to be anything clever - SQL scripts of a Q&D ETL process.
And you can get quite a lot done in a few weeks, often enough to give the appearance of significant progress.
There is a bit of an art to this. You really, really need to politically position data quality as a problem owned by the business.
So, doing the reconciliation up front, and positioning it as an analysis process give some confidence that you can get the numbers right (which is often a big driver of stakeholder buy-in). You also get the data quality issues in front of the business up front, and a much more detailed specification for the actual production ETL.
And much more certainty about implementation time lines, because you can build to a known goal. Then, everything becomes change requests :)
And, more importantly, you've got a fighting change of delivering something that the business will agree is fit for purpose, at least for some requirements, and they know that the data issues must be fixed in order to fulfil the others that are dependent on them. You can sort of push the responsibility where it belongs.
At least most of the time.
As an approach, it works OK for me.
But, it does require you to involve yourself in the analysis work from the word go, or to have business analysts who are competent enough to do it.
Essentially, it's a practical application of the GIGO principle.
@Gonsalu Oh, yes. Data warehouse projects are inherently political. You can't do them without sticking fingers in pies.
And this is where spineless project management lets the side down. If you let the project get railroaded you set it up for failure. London is full of spineless project managers.
@MarkStoreySmith That also helps, and you can alsl qualify it with some of the key assumptions, particularly where you have dependencies on third parties.
@jcolebrand, @JackDouglas I flagged this for dba.se but @casparOne just closed it. Can we re-open and migrate etc (if you think OK) stackoverflow.com/questions/8400568/…
@gbn, I just spoke with @JackDouglas and he tells me that @jcolebrand is in agreement, I'll shuffle anything over that is relatively new (answered, accepted, or not) that you flag for migration over to DBA. If it's older and has an accepted answer, I might have to confer, if it borders on programming and isn't clear that it's strictly a dba question
@gbn We're just trying to streamline (with dba at least, and thanks to @JackDouglas for having the conversation with me) that migration process. As long as we're all in agreement on who gets auto-migrate status and on what, I think we're good and we can make life easier for all.
Short, light coloured hair. I will have a rather grubby, dark green back pack and a black anorak that looks like it's seen better days. Hmmm. Not helping much, am I :)
I have a temporary number as well, until I switch providers.
For the next few days. My other one is dying; the backlight on the screen dies when I open it, so I can't use it for texting. You can call me on either for the moment, but the one in the text is my permanent one once I've shifted over.
@MarkStoreySmith Bear in mind that I can receive messages at that number but I will send from my other phone. I will text you something from it. At some point I'll move the phone number over to my other account and text you with something to that effect.
At one point I googled myself and that posting came up at the top of the search. I decided that it wasn't really the right first impression to give to recriuters or potential employers, so I changed my SO user ID to be not so personally identifiying.
Given that I do contract work I do job interview cycles on a semi-regular basis.
The Joel Test and Field work in non-IT specialised organisations
For a software company or an in-house development shop the Joel Test (as various other posters have discussed) is quite a good start. However, as a contractor one tends to find oneself working in companies that are really not geared to develop software (otherwise why would they need to hire contractors?). Since I've been working in London I don't think I've seen a company that would rate more than 3 or 4 on this scale. Usually they can get specifications, source control (even if it's just VSS) and someone to do testing. Somet…
The title of the question was "dealbreakers for new programming jobs"
I think it might be applicable for a software firm. In practice, most client sites will be pretty crap by those standards, and that tends to be where contract work is found.
Most vendors can't actually afford to hire contractors on an ongoing basis.
So, I think contracting in the city is as much the art of getting something done in suboptimal conditions as anything else.
On a tangent, if we were to expand DBA to include B.I. and database programming, do you think the SO mods could be induced to not get their backs up if we started poaching SQL questions?
@ConcernedOfTunbridgeWells oh, advanced SQL is definitely on-topic here - it is just a question whether it is also on-topic on SO. we don't really want CRUD here though
@JackDouglas Well, I'm thinking more in terms of formally closing the proposal and making it explicit that DBA (or whatever it comes to be called) covers this material.
It's got as far as commitment, so I guess the idea has some support at least. However, I think that including it into DBA's remit and explicitly stating this in the terms of DBA is probably the way to go.
Particularly if DBA has support from the powers that be at stackexchange.
@JackDouglas Give the committers a couple of weeks to respond, I reckon. If I don't get any objections from that community, we can roll the sites together and I'll get the Area51 mods to close the proposal.
In the mean time, do you think there's a general consensus within DBA that this is the right thing to do?
@gbn I'm broadly of the same view, as I think a database site would cure the problems I found with advanced B.I. questions getting lost on SO/SF. That's the main reason I proposed the B.I. site in the first place, and I think that DBA would achieve the goals I had in mind for B.I.
Especially if the SO mod community were happy to go along with it.
Is anyone else underwhelmed by all this PASS stuff (in my RSS feeds)? Or is it me being British on the wrong side of the pond (and not even in Britain)?
@JackDouglas Another thought for something that might be brought into scope - particularly in light of the lab - could be storage and HW design for database servers.
@NickChammas To be fair, I have Strunk and White on Stackoverflow.
Many times I look over some questions and answers and see the following scenarios:
Places where spelling, grammar, punctuation, syntax, and format could be improved.
Tags that are synonyms only. Based tags not used
Example : Question Tagged sql-server-2008 but not with sql-server tag
Example :...
we had a few posts about generating or interpolating series of data; e.g. I have a sales record for February and a record for December, and I want to generate records for all the months in between and give them a 0 sales number
anybody remember where we discussed that on the site?
The bare bones is this
Generate 52 week ranges
Cross join with Franchise
LEFT JOIN the actual date
ISNULL to substitute zeroes
So, like this, untested
;WITH cDATE AS
(
SELECT
CAST('20100101' AS date /*smalldatetime*/) AS StartOfWeek,
CAST('20100101' AS date /*smalldateti...
Many times I look over some questions and answers and see the following scenarios:
Places where spelling, grammar, punctuation, syntax, and format could be improved.
Tags that are synonyms only. Based tags not used
Example : Question Tagged sql-server-2008 but not with sql-server tag
Example :...
The bare bones is this
Generate 52 week ranges
Cross join with Franchise
LEFT JOIN the actual date
ISNULL to substitute zeroes
So, like this, untested
;WITH cDATE AS
(
SELECT
CAST('20100101' AS date /*smalldatetime*/) AS StartOfWeek,
CAST('20100101' AS date /*smalldateti...