Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Willfor]

Hm. Is there a profiler type thing for Python?

cProfile.

The way I get it to work on my programs is to write a separate python file with this in it:

Code: [Select]

import cProfile
cProfile.run('import [entry file name]')

[Orange Wizard]

Beautiful. Thanks.

[Squeegy]

What's your guys' favorite programming term? The one word that's the coolest piece of programmer jargon.

[Putnam]

Deep Magic/Heavy Wizardry.

Related, but not the same.

[Thief^]

Swizzle, meaning to rearrange (typically color/vector components in graphics programming). It's just a fun word

[Gatleos]

AUGH

Is there a less time and space consuming way to use multiple aggregate functions together in SQL? Something like MAX(COUNT(*)) won't work, so you have to write a subquery returning the list of grouped COUNT values, then wrap a query around it that returns MAX. But wait, I don't want the MAX value itself, I want the value in another column whose COUNT is equal to that MAX value. So now, I have to copy/paste the exact same query I already wrote to get the same table, and compare the MAX value to each row in a HAVING clause to find the highest value and display the associated value I want. And THEN I have to use that value in another nested query layer, so I copy/paste it again.

This problem only contained a couple steps, and the query is 22 lines long. Is SQL code actually this bloated, or am I missing some extremely useful feature of the language?

[karhell]

Yeah, SQL tends to get very verbose when it comes to that kind of problem. There might be something shorter than what you did, but without knowing the specifics of the problem, it's pretty much impossible to tell.

Also, is this tied to a specific DBMS ? I'm unsure about others, but oracle has some pretty nifty analytical functions that might be able to help (such as row_number() over(partition by ... order by ...) ).

[Reelya]

Well, 22 lines to do that. I'm sure Java or C++ would be many more than 22.  Though if there are repeated patterns as you suggest I'm fairly sure there would be ways to cull it down though.

[Tylui]

ptw! can't believe I didn't know about this thread.

I had a similar SQL issue the other day. A client wanted some stats on their orders, related to retailers and sales rep, but the stats they wanted were extremely complex and I spent a couple hours building this huge SQL query. I went back to make sure I solved their problem when I realized what he wanted and what he was asking for were two different things... I went back had just a nice select distinct from where order by statement. SQL is easily one of the most frustrating parts of web development sometimes, but it beats a flat file "db" by miles and miles.

[MorleyDev]

I think these are usually the points where I break out LINQPad to figure out the SQL, but I'm one of those people who find it easier to turn (functional) LINQ into SQL than writing SQL straight.

[Levi]

I wish I could post some of the ridiculous sql I've had to write at work, but I don't think I'm allowed.   

Edit:  Although, I can probably post this non-ridiculous and actually quite handy function I wrote for oracle.

Code: [Select]

CREATE OR REPLACE FUNCTION concatenate_list (p_cursor IN  SYS_REFCURSOR,delimiter IN VARCHAR2)
  RETURN  VARCHAR2
IS
  l_return  VARCHAR2(32767);
  l_temp    VARCHAR2(32767);
BEGIN
  LOOP
    FETCH p_cursor
    INTO  l_temp;
    EXIT WHEN p_cursor%NOTFOUND;
    l_return := l_return || delimiter || l_temp;
  END LOOP;
  close p_cursor;
  RETURN LTRIM(l_return, delimiter);
END;

Which lets me turn a bunch of rows into a single delimited text field(which comes in handy sometimes).

Code: [Select]

select concatenate_list(cursor (select name from example_database where velociraptors>5), ', ') from dual

bob, frank, julie, boris


[Tylui]

wow bob, frank, julie, and boris have quite a few velociraptors! I hope their electrified cages are working properly... Clever girls...

[Gatleos]

I don't mind the verbosity, I just hear this screaming in the back of my head every time I fetch a table with a specific set of conditions... then SQL's syntax forces me to discard the entire thing and set it up again to do what I want. I only have a problem with that fact that the language seems to want to cram every possible problem solution into the format of a SQL statement.

One annoying thing about it to me is the arbitrary ordering for clauses. Like, in every other imperative programming language ever, the only thing you have to remember is "statements at the top execute before statements at the bottom". But in the name of making SQL "user friendly", the clauses in SQL queries are laid out in a way that resembles English grammar. Which means they aren't written in the order they execute. So now I have to remember what the actual order is when I'm creating and referring to aliases, and it's no longer as simple as a single rule.

The most annoying side effect is what I already referred to: in another language I would load a table using a set of WHERE conditions, then perform all the operations I wanted on this one table instance I created. But contorting my solution into the shape of a SQL query means I inevitably have to write redundant and inefficient queries that load it again and again. It seems like SQL syntax only works well with smaller problems, and as soon as you get even a little complex (about the time you start using joins and subqueries) it all breaks down.

But I am doing this in a classroom, where the requirements may not accurately reflect those in a real scenario. So ignore everything I just said I guess.

[Spehss _]

ptw

[gigaraptor487]

ptw