Restored from previous forum. Originally posted by geoff.
We all want a bug free language and we all have a personal wish list for language changes, additions and improvements. Trouble is, these wishes are somewhat incompatible.
We tend to disagree over priorities. Why? I assume because we have different applications. For the newcomer writing his first game or for someone starting to learn programming what does the odd bug matter? If the language changes, no problem, he can rewrite a few lines to get things working again.
But suppose you have been programming for many years, you have tens of megabytes of source code that you need to work. When a new bug appears or a change is made to the syntax what do you do? Keep using an old version that is no longer supported, or spend days changing your code and then weeks re-testing it?
Suppose you have persuaded your employer that it's a good idea to write a few applications in PureBasic. You've got them working OK, but then a new version of PureBasic arrives, how do you explain that you need to be paid to re-work code that was working perfectly before?
From some of the comments on this Forum it is obvious that some of you have little idea how reliable compilers can be. Let me give an example. Once, when I started a new job I thought I had found a bug in the Solaris C compiler that other, more experienced, programmers were using. Not one of them believed me. They thought this so unlikely they didn't even want to be interrupted to see the problem. This is not surprising, in my entire career I have never seen a mathematical error due to a language bug in a Fortran or C program.
So am I just trying to run down PureBasic? Most certainly not, it's a great language and I admire the way Fred works so hard and patiently puts up with all our comments. I'm just making the case for reliability and stability to be given a higher priority. When it comes to complex APIs and 3D I expect a few problems with the details, but there must come a time when we have cast iron confidence in the fundamentals of the language, the maths operations, expression evaluation, control statements and so on.
Bugs, changes and reliability
-
BackupUser
- PureBasic Guru
- Posts: 16777133
- Joined: Tue Apr 22, 2003 7:42 pm
-
BackupUser
- PureBasic Guru
- Posts: 16777133
- Joined: Tue Apr 22, 2003 7:42 pm
Restored from previous forum. Originally posted by Jan.
Well, that´s certainly true. But ss it´s always good to take a look at both sides of a problem, here´s the other point of view: On the other hand there was always a workarround for these problems (I remember the time when even /2 didn´t work) and we update large sourcecodes of more than 20.000 lines each time a new version is out - it never took more than a few hours. And we were much more happy about the new possibilities than angry about the extra work. If PureBasic stops it´s internal evolution and just adds more commands that may be great for some time, but one day a new programming language will arise, build in a new, more modern fashion, and PureBasic will be history like Fortran... (thanks for this example
). Fortran may be reliable enought to program a bank system or a moon rocket, but it sucks to use it for apps or games. If PureBasic wants to go for industrial use, the one were reliability is REALLY needed, it first needs doubles - and therfore an internal rebuild
Last but not least: Wasn´t it Visual C++ which had more than 30.000 known bugs (even if not in the math part)? Complex software always has bugs - Solaris C has no bugs, as it can´t do much without external libs etc, which are - or can be - buggy. Although it´s hard to believe that there´s bugfree software from the developer of Java...
Please note: I am the last who doesn´t like a stabil and reliable language - I want it! I just wanted to show the over point of view.
Well, that´s certainly true. But ss it´s always good to take a look at both sides of a problem, here´s the other point of view: On the other hand there was always a workarround for these problems (I remember the time when even /2 didn´t work) and we update large sourcecodes of more than 20.000 lines each time a new version is out - it never took more than a few hours. And we were much more happy about the new possibilities than angry about the extra work. If PureBasic stops it´s internal evolution and just adds more commands that may be great for some time, but one day a new programming language will arise, build in a new, more modern fashion, and PureBasic will be history like Fortran... (thanks for this example
). Fortran may be reliable enought to program a bank system or a moon rocket, but it sucks to use it for apps or games. If PureBasic wants to go for industrial use, the one were reliability is REALLY needed, it first needs doubles - and therfore an internal rebuild Last but not least: Wasn´t it Visual C++ which had more than 30.000 known bugs (even if not in the math part)? Complex software always has bugs - Solaris C has no bugs, as it can´t do much without external libs etc, which are - or can be - buggy. Although it´s hard to believe that there´s bugfree software from the developer of Java...
Please note: I am the last who doesn´t like a stabil and reliable language - I want it! I just wanted to show the over point of view.
-
BackupUser
- PureBasic Guru
- Posts: 16777133
- Joined: Tue Apr 22, 2003 7:42 pm
Restored from previous forum. Originally posted by ricardo.
Geoff you are right.
I think that if PB dont want to be a toy and want to be a compiler that many developers uses, the reliability is important, not only in maths but in gadgets too.
At the moment some gadgets has strange behaviors on some situations or cant be handled properly. Per example TreeVieww, WebGadget, etc.
If someone try to convince his employe or his partner to use PureBasic and they found some strange failures in gadgets... they will spit on the proposer eyes.
Like Geoff, im not trying to putdown PB, just saying my opinion.
PB is great but giving more stabilitie to gadgets could be great.
Best Regards
Ricardo
Dont cry for me Argentina...
Geoff you are right.
I think that if PB dont want to be a toy and want to be a compiler that many developers uses, the reliability is important, not only in maths but in gadgets too.
At the moment some gadgets has strange behaviors on some situations or cant be handled properly. Per example TreeVieww, WebGadget, etc.
If someone try to convince his employe or his partner to use PureBasic and they found some strange failures in gadgets... they will spit on the proposer eyes.
Like Geoff, im not trying to putdown PB, just saying my opinion.
PB is great but giving more stabilitie to gadgets could be great.
Best Regards
Ricardo
Dont cry for me Argentina...
-
BackupUser
- PureBasic Guru
- Posts: 16777133
- Joined: Tue Apr 22, 2003 7:42 pm
Restored from previous forum. Originally posted by theogott.
I agree. Thats what I wanted to say ALSO in my reply to "integer is too slow".
There are people out there who use their time in comparisons between "language 1" and "language 2" which does not make sense so long "languge 1" is in some way "in developement".
Now, WHY are those "big compilers" so bug-free ?
Of course there are many testers in the big companies.
Maybe ONE gets the language and he is responsible that the
"floating point stuff works perfectly".
Another one get the responsibility to test the "integer stuff".
Then they have to do "heavy testing", fundamentally and with big testing-programms.
They have to report the result.
Nothing "gets out" so long the product is not "Bug free".
And they test it cause IF there wouzld be a bug, it would be their NAME that would be "in question".
Testing is for "the big compiler" not something that is done "part-time" by the developer. It is done by people who do NOTHING ELSE.
Also there needs to be some knowledge about "how to get the bugs".
Testing code does not need to be "normal code" as it must check "the limits" of the product rather as well as "normal situations".
While some things can be done automatically, testing code should be independently verified from sepate running tasks.
*************************
The best time to do things is now !
I agree. Thats what I wanted to say ALSO in my reply to "integer is too slow".
There are people out there who use their time in comparisons between "language 1" and "language 2" which does not make sense so long "languge 1" is in some way "in developement".
Now, WHY are those "big compilers" so bug-free ?
Of course there are many testers in the big companies.
Maybe ONE gets the language and he is responsible that the
"floating point stuff works perfectly".
Another one get the responsibility to test the "integer stuff".
Then they have to do "heavy testing", fundamentally and with big testing-programms.
They have to report the result.
Nothing "gets out" so long the product is not "Bug free".
And they test it cause IF there wouzld be a bug, it would be their NAME that would be "in question".
Testing is for "the big compiler" not something that is done "part-time" by the developer. It is done by people who do NOTHING ELSE.
Also there needs to be some knowledge about "how to get the bugs".
Testing code does not need to be "normal code" as it must check "the limits" of the product rather as well as "normal situations".
While some things can be done automatically, testing code should be independently verified from sepate running tasks.
*************************
The best time to do things is now !