Diese Seite mit anderen teilen ...

Informationen zum Thema:
Forum:
WinDev Forum
Beiträge im Thema:
13
Erster Beitrag:
vor 7 Jahren, 6 Monaten
Letzter Beitrag:
vor 7 Jahren, 6 Monaten
Beteiligte Autoren:
Ola, Jimbo, Al, Piet van Zanten, Arie, RdJ, DarrenF, Thomas

[WD15] Fishy error "management"

Startbeitrag von Ola am 03.02.2011 15:43

[WD15] Fishy error management in WD

I have done my best to set all error checking options to the strictest possible. I want that any error stops the show right now. But maybe I still have some very well hidden setting in the "off" position, because I don't like the error "management" of Windev at all.

Errors in code are shown with very small marks, so it is not easy to note them. Why not color the whole line red or something more visible?

In the compilation error window real errors are usually at the end, so usually a lot of scrolling through stupid warnings is needed to find them.

Testing a window does NOT ALWAYS tell if there are errors, WD just waits until it runs into one, and even then it cannot always tell exactly where the error is. Finding the error is sometimes very time consuming and frustrating.

Sometimes the reported errors are totally absurd and have no connection to the actual error situation or cause. Often these false errors concern date fields and calculations with them. They cause an awful lot of unnecessary extra work to figure out. But little by little I have learned to simply ignore them: I close WD, maybe also restart Windows and clear the project's compilation directory again, before restarting WD.

And it doesn't help a bit that the error window is filled with 54 stupid warnings about errors in my templates. The errors are there on purpose, to remind me to add the necessary window-specific code to a new window when it is based on the template. For instance, in my table template I refer several times to a "Table", which isn't there, because I add it only when a new window is created. Each table reference sauses a warning. Switching these warnings off DOES NOT work. They just stay there. I would like to be able to turn off template warnings.

And today I ran into the weirdest new one: My program had been running for weeks relatively happy with an error that I had not noticed because the problem command's parameter was given by indirection. It was a faulty parameter and referred to a file when it should have referred to a window. Still everything seemed to work ok. And now, suddenly, AFTER SEVERAL WEEKS, WD starts to complain about this error! Well, better now than never.

Simply put: it's a mess.

OK, maybe it's time to go out and shovel that new snow into those ever-growing drifts. It usually helps to cool down.

best regards
Ola
http://www.pienoismalli.com

Antworten:

Hi Ola,

Sorry I can't help with your feelings about WD errors and how they are handled by WD, but why don't you turn off the Warnings?



von DarrenF - am 03.02.2011 17:09
Thanks Darren.

That is an option, but then I won't see real warnings. But maybe that is better than to watching those 54 unnecerssary warnings all the time.

best regards
Ola

von Ola - am 03.02.2011 21:10
Hi Ola,
oh how I know your words!
As soon as one wants to use Windev hard, it shows his true colors. Then it is quickly over with 10 times faster. On the contrary by many trial and error, one needs 10 times more time.

Greetings

Thomas

von Thomas - am 04.02.2011 11:18
Mhh you create the errors yourself by setting, calling things that are not there
then you are blaiming a tool for giving the errors or am I wrong here

basicly it starts with clean programming
offcourse you have errors but lot off them are 'logical' errors made by the developer and in my opinion a tool can not help you with that
how can the tool know what you intend to do.
It gives you a result you didn't expect that is often the case, but executed exactly where you asked for

not to defend pcsoft, but in my opinion a tool can only give you warnings on direct errors and as far as I know that happens (so an argument needs x vars, you get this warning, if you don't declare your vars you get this warning, etc)
other things can only been checked during interpreting or executing ok but that must be when you programm that part
so there is also a bit of work for the programmer (yes sorry) in good thinking and clean working

when I first learned programming (6 Months ad Cap Volmac 20 years ago) I learned to write out my ideas complete, think structured and desktest before running a programm
now with all those nice tools, no problems with memory, diskspace etc, we forget those things and forget also our unit and system testing (tool must or can check this) bring it live and blame the tool when the user tells us that it isn't working

ok there are a few things where I think that the errors should be detected for running and that is for example when a query is not longer in sync with the analysis you get this error when you execute the query, was nicer when this happens when generating or when this out of sync is there (but could still be no problem when the file is generated during runtime, only should then be signaled instead of mentiond as error)

but most off the time, how sad it sounds
we as programmers forget the complexity of our programms, forget the changes in other areas because we 'don't see' this connection anymore, we forget to write out what we want to do and programm by failure (when it gives an error I have to change something) instead of programming by design (first think then execute)

yes the old waterfall method was (sometimes to) long but most of the times the designs where well thought off, now we offten try to develop agile (that is so much faster ....) and forget that programms are complex things and just do what we tell them to do
then we start complaining that we have things that don't work ...

even the best tool can't replace the human way of thinking
so it is
think befor work and try to keep the most problems/errors away
or
just start programming and see what goes wrong, but you have to debug a lot more
or
RTFM, because a lot of time we don't know good enough what a statement needs or how it work and there is a bit of problem here wit for example windev, the only manual (in English) is the the manual off pcsoft and that isn't always the most userfriendly one
in certain point they expect some knowledge there that for a starting windev programmer is not there

no blaiming or judgment only my two cents

Regards,

RdJ


von RdJ - am 04.02.2011 13:26
Rdj,

I fully agree with what you say, or at least most of it.

I do a plain vanilla app. Simple menu-table-form-type of things, but a lot of them. Procedural code. HF classic. No queries except in some reports. Nothing fancy.

I hope I was a perfect programmer. But unfortunately I am not. Therefore I need to have some kind of decent error management in my programming tool, which, also unfortunately, I do not have in Windev.

I mean, for whatever's sake, WD was waiting for weeks, before it complained about this latest problem, which I thought I didn't have.

Of course there is also another possible cause to my latest problem: Windev had an internal hitch, lost its marbles and messed my code. And then found errors that it had made itself! And, unfortunately, that would not have been the first time.

So maybe it is not just me, who makes code errors. Windev is also making them for me. Ten times faster.

best regards
Ola

von Ola - am 04.02.2011 16:14

Hi Ola, you're asking too much for! I can write zillions of lines of 'perfect' code which turn out to be bullshit at runtime and will crash in no time or after some weeks, whatever. This is the point where we're showing to be worth our money. Tests, tests, tests .. indirections are not checkable by WinDev (nor by ANY other programming tool!). There's a wonderful command in WinDev: Trace(..) - it lets you check your good/bad deeds, as a programmer of course. There's a debugger which steps you through your code line by line. You can even store test-procedures and re-run them after global code changes. WinDev has lots of tools but - as it is in our trade - it can't run/use them automatically. Thank heaven they still need programmers to do the brain work! Kind regards, Guenter


von Jimbo - am 04.02.2011 18:10
And the winner of the statement of the day contest is Jimbo with

Quote
Thank heaven they still need programmers to do the brain work!




von Arie - am 04.02.2011 18:38

:cheers:


von Jimbo - am 05.02.2011 07:05
Just a minute ago, when I was starting a project test:

"Errors found in your code. Some codes will not operate at runtime. Do you want to continue?"

No, I do not. I want to fix the errors first. No errors visible in the current code I was editing.

So I check the error window: Nothing there either. Absolutely nothing.

But I had earlier switched "display warnings" off to hide those 54 warnings I already mentioned.

Just to be sure, I turn the warnings on again. Now I see the error (and the 54 warnings).

I wonder whether this can be called fishy?

Best regards
Ola

P.S. Drinking plain coffee...:-)

von Ola - am 06.02.2011 10:47
Hello Ola

Something was changed in V14 with the saving of code and windows. Prior to V14 you could alter code in a number of areas within a window or multiple windows and still search in all that changed code. From V14 on that does not happen anymore, the search does not find changed code except for a search of "Displayed code", until the window is saved and sometimes only after it is recompiled.

I find you have to save everything and then reopen the window and click in "Code" in the main menu bar and then compile the window before search works properly or errors are detected.

Our standard operating procedure is now Alt S to save everything followed by Ctrl Shift W to close everything and then re-open the windows and compile individually and then test.

Turning the warnings on and off is something else we have got into the habit of doing to make sure that it goes and gets the changes.

It is definitely slower but maybe it is safer ?

The pity of it all is that while the features get better, allowing us to keep clients happy and excited with our apps, the working environment for us is being steadily degraded.

Regards
Al

von Al - am 06.02.2011 11:07
Hi Al

Quite obviously the error management in Windev has changed, to the worse. Also its general stability has worsened to nearly sub-standard.

Just a few versions -- and only a a few thousand new features -- ago WD was rock solid; it could be kept running for MONTHS without restarting it. Currently it must be restarted at least once a day, sometimes more often depending on how much you whip it.

And it does MESS things. Bad show.

I would gladly exchange a few hundred, even a tousand, untested new features to better stability and improved quality.

best regards
Ola

von Ola - am 07.02.2011 14:31
Hi Ola,

I totally agree.
Although I learned to live with it and it still get's the job done.
But no longer 10 times faster.
From version 14 the speed of implementing new features has certainly had a major drawback on stability.
I sure hope PCsoft will find a better way of staying alive.

Regards,
Piet

von Piet van Zanten - am 07.02.2011 15:14
Zur Information:
MySnip.de hat keinen Einfluss auf die Inhalte der Beiträge. Bitte kontaktieren Sie den Administrator des Forums bei Problemen oder Löschforderungen über die Kontaktseite.
Falls die Kontaktaufnahme mit dem Administrator des Forums fehlschlägt, kontaktieren Sie uns bitte über die in unserem Impressum angegebenen Daten.