The Whiteboard

2:03 AM

@Shog9 First @Haney wants me to bring him back a hat, now you want me to bring you back some whiskey? Something like this fine?

Shog9

@jmac I have... no idea. I'll trust your judgement

@Shog9 I don't drink whiskey ~~much~~ ever. So my judgment is worthless. But yeah, sure, I can probably find a way to bring back some Japanese whiskey.

We'll see how my baggage looks after my journey to Tokyo to conferenciate Joel to the Japanese.

Shog9

@jmac if it's a burden, don't. If it's convenient, do - and we'll enjoy some together on the roof.

(But only until your judgement is truly worthless)

@Shog9 They should have it in the airport, so it shouldn't be a burden so long as I don't forget -- my memory is the weak link in this chain.

lol

user55340

2:43 AM

... the blue name invasion continues.

user55340

I wonder if amazon.co.jp one boxes...

user55340

amazon.co.jp/gp/product/B004VJPGDM - not the hat you'd wear to work. Hint.

user55340

Yep. It does.

user55340

@jmac The 'customers bought things you looked at' after clicking on that hat got weird.

@MichaelT I know, aren't they wonderful?

user55340

2:48 AM

@jmac You should see if you can get a good whiskey glass too.

Haney is getting the big unchi hat, the little unchi hat, and an unchi pointer. That way he can deputize someone.

@MichaelT ...

Hello

I was wondering if I could ask a question about designing databases here at programmers.sc

user55340

@Malky.Kid You can certainly ask... though note that the whiskey to coffee ratio is rather on the whiskey end of the spectrum.

or, if I there is a better Q and A for it here at stack exchange

user55340

2:53 AM

I understand, thank you.

user55340

The room is fairly active, just that it tends to follow the north american work day for activity.

user55340

Ask away though... can see if I can answer.

Thanks Michael

it's about the way we approach database design

user55340

@jmac search the amazon there for "27oz beer glass"

we were taught in a way that when creating the database, we identify which are the primary keys and foreign keys, and develop the program from there

2:55 AM

@MichaelT how do you find these?

(I don't really want to know)

user55340

@jmac I was looking to see what odd beer mugs they had that you could bring back since the whiskey glass was apparently out of consideration.

user55340

That particular one caught my attention.

user55340

Btw, anyone who reads this in... oh... 8h or so. DONT CLICK THOSE LINKS AT WORK.

however, we were never really taught to identify which are the primary keys themselves in the database... we just created the tables as is.
for example, we just use

create table db.tbl(pkey as integer, data1 as varchar(100));

without really defining in the system that pkey is a primary key.

user55340

@Malky.Kid Its a reasonable approach. One of the biggest challenges though going from database to code is the object relation impedance.

Object-relational impedance mismatch

2:58 AM

we just udnerstood that it was, and we made sure that when we insert data in that table, that the pkey must not have any duplicates.

user55340

The object-relational impedance mismatch is a set of conceptual and technical difficulties that are often encountered when a relational database management system (RDBMS) is being used by a program written in an object-oriented programming language or style, particularly when objects or class definitions are mapped in a straightforward way to database tables or relational schemata. The term object-relational impedance mismatch is derived from the electrical engineering term impedance matching. == MismatchesEdit == === Object-oriented conceptsEdit === ==== EncapsulationEdit ==== Object...

user55340

One philosophy of primary key is "surrogate key for everything"

user55340

(that's my employer's philosophy)

I see.. so is it bad practice that we don't define which are the primary keys and which are the foreign ones?

when creating tables?

user55340

Well, you want to say what the relations are...

user55340

3:00 AM

but its one thing to say "the primary key is the sku for the item" (natural key) or "the primary key is a surrogate key generated from a sequence, and the sku is an attribute of that table / object"

user55340

If you follow the surrogate key philosophy, identifying the natural keys is a non-issue because you don't.

I see... we kind of approach it in a sort of whiteboard design (relationships are just written in the whiteboard) and we just straightforwardly create the tables in the database, and accommodate the system to it.

user55340

You may put constraints on the columns later... but that doesn't have anything to do with keys.

user55340

Here's the key question about keys - does your primary key have any business meaning?

well, yes. They are natural keys.

sometimes no, though

user55340

3:02 AM

I tend to the surrogate key philosophy though natural has appeal at times.

I mean, it depends on the system we are designing. Sometimes we just use the natural keys provided (ie Region number, City number) etc.

sometimes, we need to make primary keys of our own

user55340

So things that appear to be natural keys can get tricky when you find someone else needs to use that same value who isn't the same person.

I see... so it's better to declare your own primary keys then?

user55340

Thats the philosophy I tend to, and the one that the DBAs where I work to religiously follow.

and, when designing the tables, to declare relationships as you create them as well?

not just in the whiteboard?

user55340

3:05 AM

Yep, but then its from one surrogate key to another table. Don't have to worry about it because its just a number.

user55340

This table has a 1 -> 0+ relationship to that table.

We have to make changes over here..

user55340

Using database design tools makes the ER diagram much easier to work with.

thank you very much for your time Michael

user55340

You maintain the ER diagram and then do forward engineering from that.

3:06 AM

we just use phpmyadmin

and type the sql syntax raw from that

user55340

For home, I use navicat navicat.com - work we use toad.

user55340

not because it's easier though.. it's just the only method we were taught so far

we were encouraged to use sql workbench though.

user55340

nod - using a tool with the ER diagram makes it easier to maintain as you are seeing the relationships there.

user55340

MySQL workbench is an excellent program.

user55340

3:08 AM

I use navicat because it is database agnostic for its data modeler.

user55340

I can push the model into sqlite, mysql, postgreql, oracle...

agnostic... meaning you can design for any SQL?

I see!

user55340

Perhaps we should use navicat as well

user55340

Its one tool amongst many. If you are purely a MySQL shop, MySQL workbench has similar features.

3:09 AM

as you design the erd, can it produce a .sql fiel as well?

I see

user55340

My former employer was a postgresql shop. Before I worked at a MySQL. For home I waver between those two as the mood strikes me (and if I want range types or not).

user55340

As to the workbench: dev.mysql.com/doc/workbench/en/wb-data-modeling.html

user55340

Here's the bit about forward engineering to a .sql file: dev.mysql.com/doc/workbench/en/…

I see. I like navicat based on the feature you just posted though

user55340

Reverse engineering a database means having the modeling program connect to the database and build the ER diagram from the table information there.

user55340

3:12 AM

There's a free trial if you want to play with it: navicat.com/download/navicat-data-modeler

oh.. navicat isn't open source?

user55340

(the reason I have navicat is that the postgresql tools sucked (at least they did 3 years ago) - and they sucked harder on a mac (I'm a mac guy))

user55340

Nope.

although, I wouldn't mind paying to support the company.

user55340

Check out the trial, see if its good, and then consider what its worth to you and if it matches the value of the product.

3:14 AM

I understand

thank you very much Michael :)

user55340

I also do have a "support the best software - whatever the source" philosophy. I use IntelliJ for my IDE at home. Navicat for databases. Tower for doing git stuff.

Is there a way I can upvote you through chat?

user55340

Heh, nope.

user55340

But feel free to stop in any time and ask.

I won't forget this :)

Hosch250

3:39 AM

Anyone want to take a look at this? programmers.stackexchange.com/questions/298918/…

I think you question should be close. It is a bit off-topic. Maybe you should consider moving it to programmers.SE. — aloisdg 15 secs ago

Spongebob Sings Black Sabbath

4:14 AM

►

SpringLearner

Hello Everybody

Newcomers to the room must watch the Spongebob video first.

SpringLearner

I had attended one interview in TCS campuss last week and during the technical round we(Interviewer and me) have discussed about the job role.

and during all my interview generally when manager asks me do you have any questions for us then I put on questions for which project I will work if selected

but this time technical interviewer has already told me about it

so do you think will it be ok if I ask same question to manager also?

@RobertHarvey youtube is blocked at my work place

is it mandatory to watch this video before putting questions here?

Questions you should ask during an interview

7:38 AM

Unfortunately this question is off-topic for CodeReview - it might be better suited on Programmers. May I suggest that you also look into a linter, possibly, depending on your language? If you're using C#, StyleCop and Resharper will do this for you. Remember to also treat all warnings as errors :-) — A Red Herring 20 secs ago

9:46 AM

I think this question is better suited to superuser.com, not the QA exchange site for programmers. — Evil Tak 59 secs ago

This seems more a question for programmers site — colidyre 30 secs ago

Q: Predicting package delivery schedules

2:07 PM

I do not completely agree. The OP presented a image describing a software architecture. This is imo a good point to give answers there. But sure, one can definitively have another opinion. But at the end this topic definitively belongs more to Programmers than to SO. The OP can ask the question there and can edit the question there to be more specific, if needed. — colidyre 32 secs ago

user55340

2:44 PM

The delivery question is an interesting problem.

user41796

Pardoning the "tell me where to start" bit that I edited out, I'd agree

user55340

This was the core business bit (along with solving tsp) at Employer^

user55340

A great deal is based on what info you have.

user55340

Time of each past delivery? Or just start and finish?

user55340

If time, then one can build a A-B takes N. A-C takes M.

user41796

2:52 PM

Hard problems seeking hacks to simplify them

user55340

Otherwise a assume an even distribution of time and keep constraining based on other days.

user55340

Data

user55340

Hah! Edit. And it is the same as previous employer.

user55340

Context for others:

user55340

0

I'm trying to solve the following problem: A delivery man has N packages to deliver, already sorted. He starts the route at start_time, he works until end_time. I have historical data of his delivery schedules. There's no geographic information between each drop point. How can I predict the d...

user55340

2:57 PM

The thing is with gps / address locations there are existing models for transit times based on ambulance data.

but if the points themselves are not constant then you need to build topographical data

Hi there! Thanks MichaelT and Glen for the interest

Well, I don't have each drop-off waypoints in order to calculate transit times.

user55340

While I wasn't in the GIS world, my former employer did exactly this.

That's what we do right now, but I think it could be improved

user55340

Do you know the order ahead of time and lat/lon of each location?

2:59 PM

Order yes, lat/lng no...

you could use the GPS tracks to build a map and then for portions of it say this is jammed at 8 AM but clear at noon

user55340

(Btw, in 8h I can do a better job of discussion)

Johan Larsson

Rust

In Rust we trust! Rust is a systems programming language focus...

6

in case you are interested in Rust

user55340

On a rather limited device now.

user55340

ncbi.nlm.nih.gov/pmc/articles/PMC4288957

3:02 PM

Thanks, reading it right now

user55340

Lat Lon from address is a service you can subscribe to.

user55340

github.com/applieddataconsultants/wheredat

you can use openstreet map data if you want it for free

Yes, but geocoding services are generally expensive. We handle about ~200k (new) addresses a month, so we can't use the free geocoding services.

user41796

200k new addresses per month?!

3:08 PM

Yup

Since this is a history question of no practical use to programmers, it's probably off-topic here. It's being discussed on Meta and will probably get closed when the crowd from Meta descends on it. It's a good, clear and answerable question unlike much of the dross that gets asked here, but it's outside the scope of what this particular SE site is supposed to be for (in my opinion). — Mark Amery 45 secs ago

Since when did List<T> have an O(n) insertion cost?

user41796

Can I ask who you work for / what product your company makes?

user41796

@RobertHarvey Traverse to end and insert. Done.

user41796

End may even be maintained as a pointer.

3:09 PM

@RobertHarvey language?

Whose idea was that? There isn't a pointer to the last node?

C#

user55340

Previous employer also subscribed to commercial geocoding. Switched twice while I was there (under a year)

isn't that a array-backed list?

user41796

@ratchetfreak It is actually, so you don't need a pointer to the last item. Array[length-1] does the trick.

user55340

Worst case is it has to copy the list. O(n)

3:10 PM

AFAIK it is. It is resized by creating a new array and doubling the number of elements.

@jlhonora how many of those are actually legit addresses?

@GlenH7 We offer a software for delivery companies. Imagine a FedEx-like system but for smaller courier companies. So each company uploads their routes every day, each one for different contacts. Each address is formatted in a different way, and may not even be valid

user55340

LinkedList add is O(1)

user41796

@RobertHarvey In some cases, you want to pre-specify the size of the list if you know it's going to get large and have a rough idea of final size

but inserting in the middle or to the front is O(n)

3:11 PM

@MichaelT so long as you're at the front or back

So Big O is always worst case? I've got something to say about certain sorting algorithms then.

user41796

@jlhonora That's cool.

user55340

(Heh - you may be competing with previous employer)

@MichaelT do you recall which geocoding service did you use?

user41796

What about hacking your problem and finding an available list mapping addresses to geo-coordinates?

3:11 PM

@Ampt or have a iterator at the index you want to insert

user55340

I think we used Bing when I was laid off.

@ratchetfreak "If I walk the list before I do the insert, I don't have to count it in my big O time!"

@MichaelT thanks!

You could have an iterator at every spot, but you're trading storage and pre-computation for speed on one insert

user55340

@ratchetfreak if you overlay an indexed skip list, access to an element is O(log n)

3:13 PM

@GlenH7 yes, that's an option; building our own geocoding-like service. We already have 2 years worth of data, about 3MM addresses. But that doesn't work for new cities, for example

@Ampt there are algorithms that would be helped with that like merging

@jlhonora or find a existing database with data

user55340

Prior to that it was mapquest

user41796

@jlhonora Others have solved this problem before was my thinking. And there are cases where you can get reasonably close for estimation purposes. An apartment building with n apartments really only needs one main lat/long coordinate.

@ratchetfreak totally, I'd forgotten about self-hosted OSM geocoding. That could be a great options

Well, in any case, I also thought it was interesting to try to solve it without GPS coordinates.

Just for the challenge

user55340

How repeated are the addresses?

user55340

3:17 PM

Again - better answers from me in 8h

@MichaelT hmm, great question, I don't know (right now). But they are highly non-standard. Sometimes for the same destination there are two different texts, so a formatting algorithm should be applied.

user55340

(One project was hospital courier - all addresses where known in advance)

user55340

Go to clinic A, pick up tests. Go to clinic B, pick up tests. Go to hospital lab, deliver tests)

@MichaelT turn on sirens and pick up pizza before it gets cold

user55340

Heh. Not ambulances.

3:22 PM

Our current approach is to assume a 20 minute delta between deliveries. It works pretty well, but maybe we could have a variable rate depending on the time of day. Bad weather is also an interesting factor (taken from the ambulance paper).

And it could also be adjusted by driver, day of week, even month of year (take into account december holidays, for example)

user55340

As you get better data for A to B where both have been delivered before, you can refine that.

be cool if you could integrate with something like waze and get super accurate time from A to B

@MichaelT yes, that's also an option we're exploring

user55340

Furthermore, if A to A' takes 5 min, A to B takes 10, you know A' to B is no more than 15

and with the GPS track you can learn the popular routes taken by drivers and know when they get jammed up

3:25 PM

@Ampt google does that, you can even get the point-to-point distance and/or travel time: developers.google.com/maps/documentation/distance-matrix/intro . Then you can plug that into a TSP-like algorithm

aaaaaaaand he nerd sniped an entire chatroom

@MichaelT was that your previous employer?

user55340

Yep.

Aha, very interesting :) . Mine is beetrack.com

user55340

Now I work in the public sector.

3:27 PM

(warning, landing page is slow)

user55340

Ha! Sig capture. Our mobile guy was always fighting with that.

Vehicle routing with OptaPlanner

:), yes, that was a hard one (I'm also the mobile/api guy, but really into optimization and algorithms in general).

user55340

You may also find : planet.jboss.org/post/…

user55340

►

Thanks! This is a nice one as well: github.com/google/or-tools

user55340

3:32 PM

(Not the tech we used - was custom written)

This one is more accurate: developers.google.com/optimization/docs/routing/tsp

Wrapping it up:

- Read the ambulance paper
- Try to geocode the addresses; using a paid service, self-hosted geocoding service (OSM) or leverage historic data. Using that, calculate transit times.

Thanks a lot for your enthusiasm! @MichaelT, looking forward to your 8h+ insights :)

user55340

I do recall some fun of geocoding in Mexico City. The less than ideal address names.

Most of our clients are located in mexico and below. Address format quality decreases with longitude :)

user55340

Btw, geocoding in Japan is a nightmare. Not only the non-latin set, but things like "fourth house past the street lamp" is a valid address in some places.

user41796

@MichaelT This Way and That Way ?

When city planners get bored

3:39 PM

wow, hahah

well, it's not THAT bad...

user55340

@GlenH7 heh. How about parking plaza? Or center way?

user41796

I preferred Circle Drive (or whatever it was)

user55340

GIS fun, could you copy that link to here?

user41796

14 hours ago, by MichaelT

user55340

(Search and linking sucks on mobile chat)

user41796

3:41 PM

Way easier on desktop

user41796

I was wrong, it was Circle Way St.

user55340

Look at where this and that intersect.

user55340

And then there is also Winding Way.

user41796

Clearly it's a bunch of folk who would have rather been out on the water than planning the city

user55340

Any Way

3:44 PM

Do you park on Parking Way?

user55340

That's Patking Place.

And Circle Way St. is almost a circle.

user41796

@ThomasOwens No, it's either North Parking Place or South Parking Place. Sheesh, what do you think that city is like?

@GlenH7 There's also Parking Way.

And Parking Way St.

WTF IS THIS MADNESS.

user41796

Circle Way looks more like a spiral, IMO

3:45 PM

THIS IS MORE WRONG THAN BOSTON.

user41796

No, there's no parking on Parking Way, it's a shortcut between This Way and That Way.

user41796

Boston has an excuse. That place? Not so much.

Somewhere in the Boston area, there's an intersection with two streets of the same name.

user55340

I wonder if That Place is a valid street name.

I really hope that was a really good photoshop or vandalism. But I've been told it wasn't. Haven't seen it myself in person, though.

user41796

3:48 PM

@ThomasOwens Are they at least separate with a Street or Place indicator?

@GlenH7 I think so, but on the signs here, it's much smaller than the text of the street name. In fact, it's absurdly hard to see from a driver's persepective in traffic and/or in the morning and evening.

user41796

One of the best places I've ever seen with street signs was Tempe, AZ.
https://www.google.com/maps/place/2011+S+Rural+Rd,+Tempe,+AZ+85282/@33.4073208,-111.9262674,3a,75y,90t/data=!3m7!1e1!3m5!1sCIYnlYExxajYccST0-QOhA!2e0!6s%2F%2Fgeo3.ggpht.com%2Fcbk%3Fpanoid%3DCIYnlYExxajYccST0-QOhA%26output%3Dthumbnail%26cb_client%3Dmaps_sv.tactile.gps%26thumb%3D2%26w%3D203%26h%3D100%26yaw%3D151.73338%26pitch%3D0!7i13312!8i6656!4m7!1m4!3m3!1s0x872b089532d2800d:0xd0be9f8294612ee0!2s2011+S+Rural+Rd,+Tempe,+AZ+85282!3b1!3m1!1s0x872b089532d2800d:0xd0be9f8294612ee0

user41796

Hopefully that pops up into street view mode

user41796

Hi @DanPichelman

[Switches to MSTest from NUnit, reduces test spinup time from 45 seconds to 3 seconds.]

Life is good.

user41796

3:55 PM

I believe we use MSTest as well

user41796

You know you're used to having a bad network & slow servers when a data call comes back quickly and your first instinct is "No, that was too fast"

Telastyn

I like MSTest way more.

user55340

Meh. Dreaded that history question showing up here.

user41796

Questions like the enum history question cause me to wonder "Who cares? What problem is being solved here?"

2

user55340

4:12 PM

It's going to quibble over what an enum is. And if lisp had it or not.

user41796

We have closed other, similar questions where the term is imprecise as being "primarily opinion based" because of the quibbling over the term at the core of the question. Do you think this would be a good case for that?

user55340

Given recent SE actions I would be tempted to let there be quibbling prior to closing as a "nope, see this doesn't work"

user55340

And I'm out of close votes.

user41796

@MichaelT That's a fair point

user41796

As there's an implied "be a little bit more fun" behind all of that commentary.

user55340

4:20 PM

Though I'm going to vote to keep it off hot questions if it starts going that way.

user55340

And there is also a "if you think standards should be relaxed, reopen it on SO first where it got a fgitw answer)

user55340

See? Lisp.

4:58 PM

There are many other sites in the stack exchange network, some of which may consider this question on-topic. I'm not overly familiar with the policies of Programmers.SE, but this may be on topic there. Every SE site has their own community rules and policies, so it's always best to review their help pages and look around to get aquainted with what is typically accepted before posting questions. — Steven Doggart 7 secs ago

@StevenDoggart - This question would be quickly closed and likely heavily downvoted on Programmers. It's off-topic for there and appears to be too broad for all of the SE sites. — GlenH7 54 secs ago

5:19 PM

"So, durron, we've talked about a lot of goals for Q4. Do you think you'll have time to do them all?" cc @GlenH7 @enderland

user41796

Erm.... Gonna go look for a video clip of Lady Gaga's Poker Face.

@durron597 ABSOLUTELY! SIGN. ME. UP.

user41796

@durron597 How did the rest of the review go

also... former boss (from software dev) basically told me "hey you want a job here? going to be posting two positions soon" :(

user41796

@enderland How would those work with your Dec plans?

5:21 PM

so that is kind of a bummer

@GlenH7 well, taking a different position after 5 months would not be great internally

user41796

The impression I got was that quite a bit gets turned upside down for you in Dec with moving / whatnot

Yes, it does, since I'm relocating away

@GlenH7 Mostly fine, they were happier with my Q3 than I was. I told them my main goal for Q4 is training our new guy.

user41796

That's a very valid statement.

Probably for the best regardless of your plans, tbh :)

5:25 PM

@GlenH7 Honestly, it would still be my main goal no matter what happens with this background check, so it's not even dishonest

user41796

We have a running joke around here when business owners ask about when feature XYZ is due. We tell them a date (usually the next upcoming holiday) but never give it with a year.

@GlenH7 honestly it's primarily unfortunate timing...

user41796

@durron597 That's the best part of it

user41796

Completely true, nothing dishonest

@GlenH7 Well, the dishonest part of it was when were talking about other goals for the quarter, lol

I spent the vast majority of my one-on-one talking about training, improving workflow, that sort of thing

user41796

5:27 PM

... Keeps redirecting back to the part about training the new guy ...

At one point I talked about not having fixed deadline and fixed scope and then told my boss we'd reevaluate at the end of the month

user41796

WUT? I can't haz my cake and eat too?

user41796

<cue soup nazi voice> No bonus for you!

yeah...

user41796

@enderland Unfortunate timing reminds me of the time when I was directly interviewing with some of the folk in the research division of one of my old employers.

user41796

5:33 PM

Having to walk away from what had been one of my childhood dream jobs was a bit soul crushing.

Honestly though too I have a problem being content with my job, having spent several years constnatly thinking of "the next job" is hard to not want to do that now too

for that reason alone it's probably good for me to stay in my current role a while

user41796

CoL difference would have buried me though

Yeah, in this case I probably would actually have to take a small pay cut to go to that job (it's a grade lower than my currnet one)

user41796

You know you want to take a pay cut so you can switch jobs again. And then do it again in 3 months.

user41796

It's how all the cool kids climb the corporate ladder, I promise.

5:40 PM

lol

I'm amazed how few people in our higher level individual contributor IT positions have a decent IT background

like, even capable of writing basic scripts

user41796

It's about time that I added a new Progs answer....

I was just able to answer one of the bountied questions (which has a terrible accepted answer)

@GlenH7 this comment on the questino you linked is quite amusing:

I can assure you it wasn't me. I came across it in late 1995 or early 1996 while lurking on a bunch of Perl mailing lists. — Rick O Dec 1 '10 at 17:35

user41796

I need to do that with SO and build up some repz

6:35 PM

Beeraucrats.

user41796

@ThomasOwens Beer-thirty already?

user55340

New street name suggestion: North F way. To be abbreviated No F Way.

whatsisname

there is a beer thirty brand

and I almost got in a fistfight with my college roommate over its logo

user41796

@whatsisname I'm sure there wasn't any drinking leading up to that argument either

whatsisname

beer30 brand beer has caused many events in my life with good stories

user41796

6:48 PM

What was the fight over?

whatsisname

the logo

I was arguing that the second hand should be pointing down because it's "beer thirty"

but the hour hand can be wherever

and my roommate insisted that the position of the hands did not matter

so I thought the hands were badly positioned

arguing about stupid stuff + consuming significant beer 30 swill = great outcomes

user41796

At least you got a story from it

user55340

7:02 PM

Being completely sober, I will point to the slogan and suggest the hand position doesn't matter.

user41796

@MichaelT I don't think you're really in a position to weigh in on the matter if you're completely sober.

user55340

Hmm. Status completed on that Easter egg question. @ThomasOwens now has 22 non-wiki answers in the tag. I should ask 178 more so he can gold badge it.

user55340

I need 77 more discussion answers for a gold badge there.

user55340

Then I can dup them all!!!!

user55340

@gnat run a query. Extract all views on all questions. Save for a few months. Repeat. Or pull up some historical data dumps.

7:25 PM

Doesn't this SO question generally indicate a poor queuing architecture?

user41796

@ThomasOwens Not necessarily

user41796

Some MQs, like Websphere, allow the concept of a peek which allows you to skim through the messages in the queue and cherry pick the one(s) you care about

user41796

It's a different way of implementing subscribers.

Then it's not a queue anymore.

user41796

correct, it's a message queue. :-D

7:30 PM

I don't think OpenMQ or ActiveMQ implement anything like that, though.

user41796

Most MQ managers don't as it's difficult to implement. RabbitMQ doesn't do it, and I'm pretty sure Zero doesn't either.

I bet it's the big commercial ones that implement it.

user41796

The lads who worked on WMQ are definitely of a different breed. But I recall some of them saying there were many things they would do differently if they could.

user55340

They have big clients that sometimes want it and can distinguish themselves from others with unique features.

user55340

WMQ is also ancient.

user55340

7:33 PM

IBM WebSphere MQ

IBM MQ is a family of network software products that IBM launched in March 1992. It was originally called MQSeries, and was renamed WebSphere MQ in 2002 to join the suite of WebSphere products. In April 2014, it was renamed IBM MQ. IBM MQ, (often referred to as "MQ"), is IBM's Messaging solution for Enterprise and IBM's Message Oriented Middleware offering. It allows independent and potentially non-concurrent applications on a distributed system to securely communicate with each other. MQ is available on a large number of platforms (both IBM and non-IBM), including z/OS (mainframe), OS/400 (IBM...

Why not just re-submit the message if it doesn't fit the schema?

Or disable automatic deletion

(I have not experience with WebSphere, but some production experience with RabbitMQ)

user41796

The challenge is that subscribers really have zero control over what's put into a queue.

@jlhonora In Active and OpenMQ, you can't disable automatic deletion in a queue, I don't think. Once one client reads it, it won't give it to another client.

We talked about re-submitting messages, but I pointed out a whole nest of issues with that.

user41796

And in some cases, you'll have a "legacy" queue that you want to selectively tap off of.

But it may be a valid approach in this case.

user55340

7:36 PM

Queue manager applications would constantly be resubmitting.

Something that would be nice in OpenMQ, though, that I have seen, is the concept of priority.

I forget if it's not there or if there's something you need to do when you set up the queues to use it.