|
try del /F /S /Q . or rmdir /S /Q <project dir="">
Then rewrite everything from scratch ...
Espen Harlinn
Chief Architect - Powel AS
Projects promoting programming in "natural language" are intrinsically doomed to fail. Edsger W.Dijkstra
|
|
|
|
|
|
BruceN wrote: return code 97
Isn't that better than this?
#define NINETY_SIX 97
Such code is not Martian, but Earthian.
|
|
|
|
|
BruceN wrote: Martians
Matt Damon can code? Who knew.
BruceN wrote: You'd have to be a floating database guru clad in a white toga and ghandi level of sereneness to fix this goddamn clusterfuck.
I am stealing this one as my signature here.
|
|
|
|
|
One thing I have done in the past when working with legacy code is the following:
Ask for a week of no disturbance from others - preferably get an office to yourself or if you can work from home.
Then print off all the code you are working on.
Lay the code on the floor or table, if you have a big table, and slowly go through it with a pen adding comments as you read through it.
I have found this to be very effective in understanding and fixing big coding problems that I did not create(and sometimes for those that I did create).
Think of it this way - it can take you a month, or more, to fix it in your normal work time or a week of dedicated time in an office on your own - only a dumb boss would not see the benefit of allowing you the space to go through this on your own.
“That which can be asserted without evidence, can be dismissed without evidence.”
― Christopher Hitchens
|
|
|
|
|
GuyThiebaut wrote: only a dumb boss
There's always a catch.
|
|
|
|
|
|
Had problem like that--"legacy" web app written in some third party take on ASP.NET and JSP. It was "legacy" in that the code was written before we took over a contract by another contractor (so no going back to them without major $$), but it was only a year old.
Now that we were responsible, the users were asking that we fix the weird issue of extremely poor performance and daily lockups of the entire system.
I was given the task of figuring out how the code worked so I could go through and fix it. When I finally get the source code installed and 3rd party IDE up..."Hmmm...almost zero logic in the code as far as I can tell...these are strange calls..."
I go look in the database, and voila! All stored procedures. I tell myself, "Okay, I can work with this!", and start scrolling through a stored procedure...
...and scrolling...
...and scrolling...
...and scrolling...
Oh My Elephanting God! One stored procedure (the one that was suspected in causing the lockup) was 12,000 lines! And there were 30 more! Tried printing it out to do a manual refactoring--it was 250 pages of stored procedure SQL, and there was no consistency in anything.
My final report back was this:
1. This is a horrible app.
2. If you want it "fixed", I need 6 weeks and the option to just rewrite the whole thing in a more stable and common language (.NET, JSP, assembler, whatever...I didn't care--anything would be better than this nightmarish blend of old ASP and Black Speech).
3. If you want it to run "better", move it to a big iron database (we had several) and don't look at it until you want to rewrite it.
They went with option #3. Every 6-8 months, I would be told to take another look, and occasionally I would be able to make a small modification that wouldn't crash the whole system.
I left a few years later, and as far as I know, it's still out there, a monument to both Horrible Implementation and "The problem I couldn't crack".
vuolsi così colà dove si puote
ciò che si vuole, e più non dimandare
--The answer to Minos and any question of "Why are we doing it this way?"
|
|
|
|
|
Thanks for that!
12,000 line of code in one stored procedure - I don't think that pachyderm goes far enough in describing how horrendous that must have been.
David Days wrote: no consistency in anything
- that's probably the worst part, although anyone who writes a 12,000 line stored procedure is unlikely to be very well organised.
Glad they listened to you.
“That which can be asserted without evidence, can be dismissed without evidence.”
― Christopher Hitchens
|
|
|
|
|
Sounds more it has been written by some consultant company who deliver something and then run away. Close to Martians but much more greedy. Know this too good and I absolutely feel with you.
If I cannot understand how something works within half a day of studying it, I write it new. This isn't always easy or possible but if I can, I do it. Saves me and the company a lot of trouble and reduces risk.
|
|
|
|
|
BruceN wrote: So bizarre, so obtuse you just can't see the answer?
I've been with the same company for 20 years now, and I've seen this many times. You know the something even worse? When you realize that you wrote the bizarre, obtuse code in question...
|
|
|
|
|
|
Congratulations, you've learned a new skill...Real-World Programming. Sorta like Real-Time Programming, except someone else wrote all the code, nothing ever works, so it doesn't matter if the tasks finish within the allotted time slot.
Any Real-World programmer knows that return 97 means error 96 with one argument, what's the problem? And yes, that print routine should have been one base class with 23 levels of inheritance piled on, sloppy programming there, sure to bring down your cyclomatic complexity score.
Set some realistic goals and all will be well. Myself, I'm working on getting that cyclomatic complexity score to overflow and crash Visual Studio.
|
|
|
|
|
Bruce,
You have to be the Yoda! What they did is done. You "get" to clean it up.
You "get" to use that brain to figure out what brainless people were doing.
You get to be the white hat, and the hero!
Everyday, just make something better... And you will leave a legacy!
In this case, the code is the stimulus. And the most amazing part of being human,
is we get to CHOOSE our response to stimuli... We can get all upset. Lose hope/faith.
Cry that it isn't fair.
Or, we can take some of these examples, put them aside for a future article/book/training
material (as in... If you do this while working for me, you will be fired, and I will publicly
post what you did!).
And then clean it up. The hardest thing is to be the Jedi that only makes the changes
that are required while making things a little better each time. (Most programmers see horrible
stuff and want to re-write all of it. That is far easier than understanding illogical code, and
fixing it with the smallest, best change possible).
Finally... Think of ways that would prevent this in the future. Code Reviews!!!
Start having them. Stop the creation of bad code NOW.
When I approach it this way, I find it empowering.
Otherwise, if your perspective is off, one could reduce fixing "well written" code to "All I do is fix bugs or implement STUPID business changes that they should have put in there already". Which might be accurate, but does nothing to answer the question "How can I become Happier?"
|
|
|
|
|
Rather than rant, you could just do like me and my coworker did once -- named an award of coding shame after the former engineer and passed it around for years whenever we caught each other, or ourselves, doing something so stupid.
We can program with only 1's, but if all you've got are zeros, you've got nothing.
|
|
|
|
|
I feel for you. Probably the worst code I ever worked on was a real-time program where the apparently nonsensical sequence of events had a real-world impact that was not immediately obvious. 
Fletcher Glenn
|
|
|
|
|
Try a project written by a sucksession of contractors all of whom codded differently badly with simple aspx pages with 2k+ line code behinds. The contractors are long gone. Then anyone with even passing knowledge of how the application works doesn't have time to waste on explaining fertilizer ( think organic) to you. Can you get this done by tomorrow?
|
|
|
|
|
Hmmm... If you know what the program's supposed to do, then I'd advise you to delete it and write a new one... by the time you figure out how this crap works, you might as well have done that anyway.
|
|
|
|
|
You didn't say how legacy "legacy" is, but there are always logical reasons for these things.
The first reason is that someone was making changes and didn't get around to making the change everywhere.
The meaning of the magic code 97 can be found on a note pasted under the system operator's keyboard.
The print routine call is a masterpiece of software engineering that allows you to call different routines with the same code.
You have a remarkably stable database to be able to stand firm for a thousand years. I would be grateful.
The toga might be allowed on some casual Fridays. I would check into it.
|
|
|
|
|
That's my code you're complaining about.  Seriously though, I've run into code like this and you're almost better off rewriting from scratch when it's this bad.
|
|
|
|
|
Sounds like the problems I face on a day to day basis...
What I try to do is first of all figure out what is required of me. Then, very carefully inject what I need to do somewhere that looks appropriate. The key is to move things around but keep what is does and when it does it.
"Program testing can be used to show the presence of bugs, but never to show their absence."
<< please vote!! >></div>
|
|
|
|
|
|
Moving from "old school" GWT to UI Binder is not "getting current".
Elephant elephant elephant, sunshine sunshine sunshine
|
|
|
|
|
It can be... I mean, trying to use UIBinder with generic classes kinda feels like getting a large amount of current shunted through your skull...
|
|
|
|
|
I'm shocked that you would say such a thing.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
