Re: [fpc-pascal] with statement using mulltiple objects
2014-09-17 3:11 GMT+02:00 Flávio Etrusco flavio.etru...@gmail.com: On Tue, Sep 16, 2014 at 8:09 PM, Philippe phili...@quarta.com.br wrote: on compiler implementation: I read in this list that the group lost months due a with problem. People use with, and I guess lot of people use it! I checked the reference http://www.freepascal.org/docs-html/ref/refsu58.html#x155-16500013.2.8 and there is not any warning ... just: The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; without any warning ... (...) Why would it? As people replied in your thread, this has always been the behavior (at least in TP and Delphi), so if anything the only thing lacking is some minor documentation. Also please don't hijack threads. I guess different communities have different rules, but I don't understand (and I'd better understand if I don't want to do the same mistake): in what way was Philippe's post hijacking? This question goes to Philippe too, as he wrote we may be out of subject which shows he felt something wrong. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Wed, Sep 17, 2014 at 4:38 AM, Frederic Da Vitoria davito...@gmail.com wrote: 2014-09-17 3:11 GMT+02:00 Flávio Etrusco flavio.etru...@gmail.com: On Tue, Sep 16, 2014 at 8:09 PM, Philippe phili...@quarta.com.br wrote: on compiler implementation: I read in this list that the group lost months due a with problem. People use with, and I guess lot of people use it! I checked the reference http://www.freepascal.org/docs-html/ref/refsu58.html#x155-16500013.2.8 and there is not any warning ... just: The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; without any warning ... (...) Why would it? As people replied in your thread, this has always been the behavior (at least in TP and Delphi), so if anything the only thing lacking is some minor documentation. Also please don't hijack threads. I guess different communities have different rules, but I don't understand (and I'd better understand if I don't want to do the same mistake): in what way was Philippe's post hijacking? This question goes to Philippe too, as he wrote we may be out of subject which shows he felt something wrong. -- Frederic Da Vitoria (davitof) The mistake was actually mine :-$ I had deleted part of the thread and didn't realize this was the old thread. Sorry Phillippe, and sorry for the noise everybody. -Flávio ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
2014-09-14 16:23 GMT+02:00 Mark Morgan Lloyd markmll.fpc-pas...@telemetry.co.uk: I'd be far happier if there were provision for declaring a temporary shortcut symbol: with shortcut: TSomething= TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin shortcut.DirLogged := true; shortcut.DirHatFocus := false; shortcut.SubDirsExpanded := true; Yes, something close to alias in SQL, instead of plain elision which with currently performs. This would allow for a clearer writing without the pitfalls of the Pascal with. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? Em 16.09.2014 05:44, Frederic Da Vitoria escreveu: 2014-09-14 16:23 GMT+02:00 Mark Morgan Lloyd markmll.fpc-pas...@telemetry.co.uk: I'd be far happier if there were provision for declaring a temporary shortcut symbol: with shortcut: TSomething= TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin shortcut.DirLogged := true; shortcut.DirHatFocus := false; shortcut.SubDirsExpanded := true; Yes, something close to alias in SQL, instead of plain elision which with currently performs. This would allow for a clearer writing without the pitfalls of the Pascal with. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org [2] ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [1] Links: -- [1] http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [2] http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
2014-09-16 12:26 GMT+02:00 Philippe phili...@quarta.com.br: Em 16.09.2014 05:44, Frederic Da Vitoria escreveu: 2014-09-14 16:23 GMT+02:00 Mark Morgan Lloyd markmll.fpc-pas...@telemetry.co.uk: I'd be far happier if there were provision for declaring a temporary shortcut symbol: with shortcut: TSomething= TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin shortcut.DirLogged := true; shortcut.DirHatFocus := false; shortcut.SubDirsExpanded := true; Yes, something close to alias in SQL, instead of plain elision which with currently performs. This would allow for a clearer writing without the pitfalls of the Pascal with. could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? I don't think it would make it safe, only slightly safer. Imagine: you are using 2 nested with which have a field with the same name. The compiler would tell you so. You'd check and decide that it's OK because you really intended to mean the internal with. Then, later, you remove or rename the field which is used in the internal with. The compiler won't send you any warning since you will have removed a danger. But the meaning of your code will have changed and will now be wrong. Will you remember that there was a warning before and will you realize that the fact this warning disappeared means trouble? -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Tue, 16 Sep 2014 07:26:24 -0300 Philippe phili...@quarta.com.br wrote: could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? How to get rid of the hint? Mattias ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
2014-09-16 12:36 GMT+02:00 Frederic Da Vitoria davito...@gmail.com: 2014-09-16 12:26 GMT+02:00 Philippe phili...@quarta.com.br: Em 16.09.2014 05:44, Frederic Da Vitoria escreveu: 2014-09-14 16:23 GMT+02:00 Mark Morgan Lloyd markmll.fpc-pas...@telemetry.co.uk: I'd be far happier if there were provision for declaring a temporary shortcut symbol: with shortcut: TSomething= TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin shortcut.DirLogged := true; shortcut.DirHatFocus := false; shortcut.SubDirsExpanded := true; Yes, something close to alias in SQL, instead of plain elision which with currently performs. This would allow for a clearer writing without the pitfalls of the Pascal with. could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? I don't think it would make it safe, only slightly safer. Imagine: you are using 2 nested with which have a field with the same name. The compiler would tell you so. You'd check and decide that it's OK because you really intended to mean the internal with. Then, later, you remove or rename the field which is used in the internal with. The compiler won't send you any warning since you will have removed a danger. But the meaning of your code will have changed and will now be wrong. Will you remember that there was a warning before and will you realize that the fact this warning disappeared means trouble? Your suggestion would be safe if the option was configured to error. Then, in such a situation, the developer would have to fully qualify each ambiguous field which should remove any chance of error. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Il 16/09/2014 12:26, Philippe ha scritto: could the compiler avoid with pitfalls? The with construct can be very handy, but should be used with caution, like a lot of other constructs, declarations, etc. E.g. for symbols present with the same name in different units, depending on the order of the uses declaration the one or the other will be picked up. The IDE codetools provide already a powerful checking tool: you stop with the cursor on a symbol, and it tells you which reference will be taken. More than that would be an overkill: if you don't take the time to verify that your with does what you expect, you can't blame neither the construct nor the compiler. Just my 5 cents Giuliano ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 16.09.2014 12:37 schrieb Mattias Gaertner nc-gaert...@netcologne.de: On Tue, 16 Sep 2014 07:26:24 -0300 Philippe phili...@quarta.com.br wrote: could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? How to get rid of the hint? Either get rid of the with or use a compiler directive to disable it ;) Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
2014-09-16 14:42 GMT+02:00 Sven Barth pascaldra...@googlemail.com: Am 16.09.2014 12:37 schrieb Mattias Gaertner nc-gaert...@netcologne.de : On Tue, 16 Sep 2014 07:26:24 -0300 Philippe phili...@quarta.com.br wrote: could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? How to get rid of the hint? Either get rid of the with or use a compiler directive to disable it ;) Regards, Sven ... or rewrite the offending reference so that the compiler won't complain. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
I imagine that who really want to get rid from the hint ... don´t use Pascal anyway! or do have strong test tools!!! one point is that (many?) people use with ... and compiler can make it safer. may be as described Frederic Da Vitoria ... producing error when may be confused. (but let configure it!) Philippe Em 16.09.2014 09:48, Frederic Da Vitoria escreveu: 2014-09-16 14:42 GMT+02:00 Sven Barth pascaldra...@googlemail.com: Am 16.09.2014 12:37 schrieb Mattias Gaertner nc-gaert...@netcologne.de: On Tue, 16 Sep 2014 07:26:24 -0300 Philippe phili...@quarta.com.br wrote: could the compiler avoid with pitfalls? now the compiler attach a property to the closest with where it finds it. the compiler could check if the property appears in another with of the with stack and produce a message (hint, warning or error, which may be configurated). is not it a way to make the with safe? How to get rid of the hint? Either get rid of the with or use a compiler directive to disable it ;) Regards, Sven ... or rewrite the offending reference so that the compiler won't complain. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org [2] ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [1] Links: -- [1] http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [2] http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 09/16/2014 03:41 PM, Philippe wrote: I imagine that who really want to get rid from the hint ... don´t use Pascal anyway! or do have strong test tools!!! one point is that (many?) people use with ... and compiler can make it safer. may be as described Frederic Da Vitoria ... producing error when may be confused. (but let configure it!) The issue is that when you introduce this hint, it is almost certain that someone will ask `how to disable that hint?` on this list. IMO it would be a waste of time to implement such a thing. If I were a developer (hmm, actually I am one :-) ), and I would work on a project like FPC (meaning: in my free time, not getting paid to do the job), I would invest my time in something that actually means something, like fixing a bug, optimizing the RTL for a specific target, writing documentation, etc... We are programmers after all, not end users who want to be treated with silken gloves. You've only got 10 choices really: either you like the `with` statement, or you don't. In the latter case don't use it. Remember that the original question was about functionality, what this thread has now evolved into is, in my opinion, off-topic. -- Ewald ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Mon, Sep 15, 2014 at 2:55 AM, Sven Barth pascaldra...@googlemail.com wrote: On 14.09.2014 18:05, Philippe wrote: Take this example: === code begin === (...) procedure TestWith; var p: PTest; begin New(p); with p^ do begin Prop1 := 42; Prop2 := 21; end; Dispose(p); end; procedure TestWithout; var p: PTest; begin New(p); p^.Prop1 := 42; p^.Prop2 := 21; Dispose(p); end; begin end. === code end === This is the relevant code generated for TestWith: === asm begin === (...) === asm end === As you can see the expression p^ is only evaluated once in the TestWith case while it's evaluated twice in the TestWithout one. So it's only minimally faster in this example (one less memory access), but if you use enough members of TTest it a more or less tight loop it might even be noticeable. Regards, Sven I expected FPC would optimize this (with just -O1) :-( -Flávio ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
on compiler implementation: I read in this list that the group lost months due a with problem. People use with, and I guess lot of people use it! I checked the reference http://www.freepascal.org/docs-html/ref/refsu58.html#x155-16500013.2.8 and there is not any warning ... just: The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; without any warning ... (and there is noting about the syntax with xxx[ nn]^ = A do ... describe by someone earlier) I am not telling that an implementation in the compiler should be of higher priority!!! I appreciate a lot the great job that has been done, and is still going on! I know I won't be able to do it!!! I'ld really like to contribute ... but ... my age and qualification ... won't help very much!!! But I am pretty sure that this implementation could have saved many time to many people (as one of the developper wrote) ... and still may save. But as it has been written ... we may be out of subject! Philippe Em 16.09.2014 11:08, Ewald escreveu: On 09/16/2014 03:41 PM, Philippe wrote: I imagine that who really want to get rid from the hint ... don´t use Pascal anyway! or do have strong test tools!!! one point is that (many?) people use with ... and compiler can make it safer. may be as described Frederic Da Vitoria ... producing error when may be confused. (but let configure it!) The issue is that when you introduce this hint, it is almost certain that someone will ask `how to disable that hint?` on this list. IMO it would be a waste of time to implement such a thing. If I were a developer (hmm, actually I am one :-) ), and I would work on a project like FPC (meaning: in my free time, not getting paid to do the job), I would invest my time in something that actually means something, like fixing a bug, optimizing the RTL for a specific target, writing documentation, etc... We are programmers after all, not end users who want to be treated with silken gloves. You've only got 10 choices really: either you like the `with` statement, or you don't. In the latter case don't use it. Remember that the original question was about functionality, what this thread has now evolved into is, in my opinion, off-topic. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Tue, Sep 16, 2014 at 8:09 PM, Philippe phili...@quarta.com.br wrote: on compiler implementation: I read in this list that the group lost months due a with problem. People use with, and I guess lot of people use it! I checked the reference http://www.freepascal.org/docs-html/ref/refsu58.html#x155-16500013.2.8 and there is not any warning ... just: The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; without any warning ... (...) Why would it? As people replied in your thread, this has always been the behavior (at least in TP and Delphi), so if anything the only thing lacking is some minor documentation. Also please don't hijack threads. -Flávio ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Monday 15 September 2014 07:55:20 Sven Barth wrote: As you can see the expression p^ is only evaluated once in the TestWith case while it's evaluated twice in the TestWithout one. So it's only minimally faster in this example (one less memory access), but if you use enough members of TTest it a more or less tight loop it might even be noticeable. Or if the reference is calculated for example in an array of record or a getter function. Martin ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Mon, September 15, 2014 07:55, Sven Barth wrote: On 14.09.2014 18:05, Philippe wrote: someone wrote about a better performance using with. is that true? even with a simple pointer as in: . . This is the relevant code generated for TestWith: === asm begin === # [28] with p^ do begin movl-4(%ebp),%ebx # [29] Prop1 := 42; movl%ebx,%eax movw$42,%dx callP$TWITHTEST_TTEST_$__SETPROP1$SMALLINT # [30] Prop2 := 21; movl%ebx,%eax movw$21,%dx callP$TWITHTEST_TTEST_$__SETPROP2$SMALLINT === asm end === and this for TestWithout: === asm begin === # [42] p^.Prop1 := 42; movl-4(%ebp),%eax movw$42,%dx callP$TWITHTEST_TTEST_$__SETPROP1$SMALLINT # [43] p^.Prop2 := 21; movl-4(%ebp),%eax movw$21,%dx callP$TWITHTEST_TTEST_$__SETPROP2$SMALLINT === asm end === As you can see the expression p^ is only evaluated once in the TestWith case while it's evaluated twice in the TestWithout one. So it's only minimally faster in this example (one less memory access), but if you use enough members of TTest it a more or less tight loop it might even be noticeable. Have you tried to enable optimizations? Diff of the two code fragments generated with 2.6.4 and -O3: --- t_with_o3.s Mon Sep 15 08:59:07 2014 +++ t_without_o3.s Mon Sep 15 08:59:32 2014 @@ -1,18 +1,14 @@ -# [26] New(p); +# [40] New(p); movl$0,%eax callfpc_getmem movl%eax,%ebx -# [28] with p^ do begin - movl%ebx,%esi -# [29] Prop1 := 42; - movl%esi,%eax +# [42] p^.Prop1 := 42; movw$42,%dx callP$TWITHTEST_TTEST_$__SETPROP1$SMALLINT - movl%esi,%eax -# [t.pas] -# [30] Prop2 := 21; +# [43] p^.Prop2 := 21; + movl%ebx,%eax movw$21,%dx callP$TWITHTEST_TTEST_$__SETPROP2$SMALLINT -# [33] Dispose(p); +# [45] Dispose(p); movl%ebx,%eax callfpc_freemem Note that there are actually more instructions generated for version using 'with' in this case (obviously depending on the context). In other words, the remark about micro-optimizations (which may not necessarily lead to expected results) applies here... Tomas ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 2014-09-14 11:49, Sven Barth wrote: Though to be honest Lazarus handles with-statements in context of debugging much more gracefully than Delphi... We all know Lazarus is way better that Delphi - so that's always good to know. :) As already mentioned, human refactoring of WITH statements can, and has, caused lots of debugging issues afterwards. Two or more classes used in a with statement, both classes have the same property name - how the hell must the developer know (and the compiler for that matter) which one is meant where. Bottom line (for me), WITH gives more trouble than its worth, so I avoid it at all costs. Regards, - Graeme - -- fpGUI Toolkit - a cross-platform GUI toolkit using Free Pascal http://fpgui.sourceforge.net/ ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 2014-09-14 15:41, Vojtěch Čihák wrote: because Editor is also public property of TStringGrid. Here would be solution to rename parameter to AEditor. Yes, that is a classic issue I have often come across. Regards, - Graeme - -- fpGUI Toolkit - a cross-platform GUI toolkit using Free Pascal http://fpgui.sourceforge.net/ ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 2014-09-14 13:30, Marco van de Voort wrote: WITH doesn't, and you risk that if your class or one of its ancestor has such property, and your error goes unnoticed. +1 Not to mention that you can't use tooltip debugging when it comes to the WITH statement - at least this was the case last time I tried (I don't use WITH, so it could have been a long time ago). Regards, - Graeme - -- fpGUI Toolkit - a cross-platform GUI toolkit using Free Pascal http://fpgui.sourceforge.net/ ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Graeme Geldenhuys wrote: As already mentioned, human refactoring of WITH statements can, and has, caused lots of debugging issues afterwards. If Dijkstra had written a letter to the CACM cautioning against the With statement, I wonder whether Wirth would have headlined it With Statement Considered Harmful? [Ref: EWD-1308] -- Mark Morgan Lloyd markMLl .AT. telemetry.co .DOT. uk [Opinions above are the author's, not those of his employers or colleagues] ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Bernd Oppolzer wrote: PL/1 was designed in the mid 60s, and Wirth of course knew about PL/1, because he was at IBM at the time, and of course he was aware of the problems and pitfalls of PL/1. I didn't know that Wirth had been directly associated with IBM. I did know that while at Stanford he'd supervised Larry Breed who was writing the first fairly-complete computerised APL implementation, but my recollection is that they used FORTRAN since it was the only language that ran on the various computers that the involved parties had available. http://infolab.stanford.edu/TR/CS-TR-66-47.html -- Mark Morgan Lloyd markMLl .AT. telemetry.co .DOT. uk [Opinions above are the author's, not those of his employers or colleagues] ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 15.09.2014 09:08 schrieb Graeme Geldenhuys mailingli...@geldenhuys.co.uk: On 2014-09-14 13:30, Marco van de Voort wrote: WITH doesn't, and you risk that if your class or one of its ancestor has such property, and your error goes unnoticed. +1 Not to mention that you can't use tooltip debugging when it comes to the WITH statement - at least this was the case last time I tried (I don't use WITH, so it could have been a long time ago). With Lazarus this works, which is what I meant in my other mail ;) In Delphi this AFAIK still does not work... (at least it did not in Delphi 2007) Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Sun, Sep 14, 2014 at 02:30:01PM +0200, Marco van de Voort wrote:
In our previous episode, J?rgen Hestermann said:
-
d:=TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]; // local var d
d^.DirLogged := true;
d^.DirHatFocus := false;
d^.SubDirsExpanded := true;
[...] -
Long-winded and still not the same clarity as in the with statement.
But the important part is that it forces each property/field to be a member
of T, or get a compiler error.
WITH doesn't, and you risk that if your class or one of its ancestor has
such property, and your error goes unnoticed.
I have to believe the WITH statement is more efficient
than the C language equivalent where an extra
convenience pointer is declared as a variable and
allocated. The WITH statement allows the compiler to
generate the same code as a fully qualified variable name
without extra overhead in the form of a dummy pointer.
Just as the GOTO statement needs to be used with careful,
close locality of scope, the WITH statement should be used
when the fields will all come from the same record
definition and ideally have distinct names. The safe
approach would be to qualify all field references that are
potentially ambiguous even while using a WITH statement.
The danger of surprises in scope evaluation for a simple
and short variable name within a WITH statement scope are
not that different from the danger of unit global
variables. Each unit creates its own local name space,
which usually does the programmer a favor, but not always.
Now the order units are referenced in the USES statement
of each individual unit matters. What do you think
happens when the programmer is casual about the USES order
for helper utility units with duplicate global variable
names?
Unit FooUtil1;
var ErrorString : string;
Unit BarUtil2
var ErrorString : string;
Unit CodeSection;
uses FooUtil1, BarUtil1;
procedure Main
begin
ErrorString := ''; {Clears error string in last unit}
.
.
end;
___
fpc-pascal maillist - fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 13 September 2014 20:29, vfclists . vfcli...@gmail.com wrote: On 13 September 2014 20:02, Sven Barth pascaldra...@googlemail.com wrote: On 13.09.2014 20:13, vfclists . wrote: According to the docs the with statement when used with mulltiple objects only works with the last parameter. The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; -- I thought that if all the objects have the property the statement would apply to them eg. with Label1, Label2, Label3 do Caption = 'Text'. I thought all the 3 labels would have the caption set, but only Label3's caption is set. Is this a difference between Delphi's Object Pascal and FreePascal, or has Delphi's Object Pascal always worked in the same manner? It has always worked in this manner. It's just for abbreviating typing, not for combining properties. (and then the with-variant with multiple elements is even more seldomly used than the single-element one...) Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal What then is the nesting for if the command will only apply to the last item with the property? Is each nested element supposed to be a property of the enclosing element? What is the rationale for such a statement, ie using multliple elements? -- Frank Church === http://devblog.brahmancreations.com This is a fragment I was using with multiple elements. So it means that all this time it was only working because all of them had been set at design time and the command was only applied to the last one. I wonder where I got the impression that because they all shared the property the command applied to all of them. with qrySCParameters, qrySavedCommands, qryTemplateParameters, qryLiveCommands, qryLCParameters do begin Connection := dmCentral.localCommandsConnection; end; dmCentral.localCommandsConnection.Connected := True; I am happy with the use of 'with' on a single element, but using it with multiple elements is plainly evil. -- Frank Church === http://devblog.brahmancreations.com ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 2014-09-13 21:04, Mattias Gaertner wrote: What is the rationale for such a statement, ie using multliple elements? Less typing. ...and increases grey hair growth (due to lots more trouble debugging). :-) Regards, - Graeme - -- fpGUI Toolkit - a cross-platform GUI toolkit using Free Pascal http://fpgui.sourceforge.net/ ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 2014-09-13 21:31, Marius wrote: programmers know that and try to avoid the with statement. +1 Regards, - Graeme - ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 2014-09-13 22:31, schrieb Marius: Mattias Gaertner wrote: All seasoned programmers know that and try to avoid the with statement. Realy? I love it. Consider these two varianst of code from one of my programs: - with TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin DirLogged := true; DirHatFocus := false; SubDirsExpanded := true; NonLoggedDirsInBranch := 0; LastSubDirIndex := -1; ErrorString := ''; fillchar(StatistikOfFiles,sizeof(StatistikOfFiles),0); fillchar(StatistikOfDirs ,sizeof(StatistikOfDirs) ,0); Verzeichnis := Form1.Verzeichnisbaum.RootNode; DirName := DirectorySeparator; DirNameLowerCase := LowerCase(DirName); TRightMostPeriod := 0; DirSize := 0; TFileAttributes := 0; fillchar(TDateWritten,sizeof(TDateWritten),0); TFlagSet := []; SubDirs := nil; Files := nil; end; - and - TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.DirLogged := true; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.DirHatFocus := false; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.SubDirsExpanded := true; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.NonLoggedDirsInBranch := 0; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.LastSubDirIndex := -1; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.ErrorString := ''; fillchar(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.StatistikOfFiles,sizeof(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.StatistikOfFiles),0); fillchar(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.StatistikOfDirs ,sizeof(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.StatistikOfDirs) ,0); TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.Verzeichnis := Form1.Verzeichnisbaum.RootNode; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.DirName := DirectorySeparator; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.DirNameLowerCase := LowerCase(DirName); TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.TRightMostPeriod := 0; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.DirSize := 0; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.TFileAttributes := 0; fillchar(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.TDateWritten,sizeof(TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.TDateWritten),0); TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.TFlagSet := []; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.SubDirs := nil; TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^.Files := nil; - Which one would you prefer? The first one is *much* less error prone because it is easier to read and makes it therefore clearer than the second example. Also, the calculation of the pointer is done multiple times in the second example which makes the code slower. I love Pascal also because of the WITH statement. The risks are the same as when using multiple units which define identical named identifiers. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Sun, 14 Sep 2014 11:58:22 +0200 Jürgen Hestermann juergen.hesterm...@gmx.de wrote: Am 2014-09-13 22:31, schrieb Marius: Mattias Gaertner wrote: All seasoned programmers know that and try to avoid the with statement. The quote is from Marius, not from me. Realy? I love it. Consider these two varianst of code from one of my programs: - with TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin DirLogged := true; DirHatFocus := false; SubDirsExpanded := true; [...] - and - d:=TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]; // local var d d^.DirLogged := true; d^.DirHatFocus := false; d^.SubDirsExpanded := true; [...] - Mattias ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 2014-09-14 12:08, schrieb Mattias Gaertner: All seasoned programmers know that and try to avoid the with statement. The quote is from Marius, not from me. Correct. I deleted the wrong reference. ;-( Realy? I love it. Consider these two varianst of code from one of my programs: - with TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin DirLogged := true; DirHatFocus := false; SubDirsExpanded := true; [...] - and - d:=TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]; // local var d d^.DirLogged := true; d^.DirHatFocus := false; d^.SubDirsExpanded := true; [...] - Long-winded and still not the same clarity as in the with statement. Do you also prepent *all* function and variable names with the unit name they come from? Do you realy *always* write LazUTF8.UTF8Delete instead of UTF8Delete ? Nested WITH statements and the unit concept share the same problem but still the benefits outweigh the drawbacks IMO (once you get used to it and keep possible problems in mind). ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 14.09.2014 09:36 schrieb Graeme Geldenhuys mailingli...@geldenhuys.co.uk: On 2014-09-13 21:04, Mattias Gaertner wrote: What is the rationale for such a statement, ie using multliple elements? Less typing. ...and increases grey hair growth (due to lots more trouble debugging). :-) Though to be honest Lazarus handles with-statements in context of debugging much more gracefully than Delphi... Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Jürgen Hestermann wrote: Realy? I love it. Consider these two varianst of code from one of my programs: This is a simpler example than the one with with the 4 levels deep (which can be quite a puzzle if there are multiple records containing the same fieldname(s)). And yes you may love them but please read on.. Which one would you prefer? The second, but not for readability or typing, I would suggest a third solution by using a variable (like Matthias also showed). The point is you are very likely to run into refactoring problems. Suppose your routines are in 2 different methods in a class and there is class inheritance and one of the base class has a property errorstring. Suppose you decide to refactor the errorstring in you record structure to something else (and forget to adjust it properly which is human). Your first solution will compile but it would reset the wrong errorstring, your second solution will simply not compile because it has been prefixed (hence my preference to pick that one). You can argue its the fault of the programmer but why create an unsafe situation in the first place (and that just what the with statement does)... We have spend weeks (if not months) off hunting down this kind of bugs. And its this kind of bug what showed up frequently in teams of programmers and that made us decide to ban the with statement as much as possible. Sooner or later the with stament is going to bite you, I just hope you wont have to spend that much time on resolving it ;( Marius ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Jürgen Hestermann wrote: Do you also prepent all function and variable names with the unit name they come from? Do you realy always write LazUTF8.UTF8Delete instead of UTF8Delete ? Nested WITH statements and the unit concept share the same problem but still the benefits outweigh the drawbacks IMO (once you get used to it and keep possible problems in mind). Good point, as far as i know there is not a solid solution for units unless those global procedures like utf8delete where rewritten as class procedures. Marius ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
In our previous episode, J?rgen Hestermann said: - d:=TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]; // local var d d^.DirLogged := true; d^.DirHatFocus := false; d^.SubDirsExpanded := true; [...] - Long-winded and still not the same clarity as in the with statement. But the important part is that it forces each property/field to be a member of T, or get a compiler error. WITH doesn't, and you risk that if your class or one of its ancestor has such property, and your error goes unnoticed. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
-1. !!! I like with ... even it may give trouble times to times!!! ... but I work alone on my programs ... as always ... nothing perfect! always should consider the conditions!! Em 14.09.2014 09:30, Michael Thompson escreveu: We have spend weeks (if not months) off hunting down this kind of bugs. And its this kind of bug what showed up frequently in teams of programmers and that made us decide to ban the with statement as much as possible. Sooner or later the with stament is going to bite you, I just hope you wont have to spend that much time on resolving it ;( +1. With statement was banned at my old job as well for exactly the reasons stated (by everyone :) ) Mike ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [1] Links: -- [1] http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Hello All, My two cents. The with statement is crystal clear for me. If newbies prefer to use different manners to code the things, that are free to do it. Using some kind of intermediate variable ? That's make to code difficult to understand, need do open begin/end blocks in some situations. With make the code very clear, probably helps the compiler a lot. Well used, it's 100% safe - I *never* encountered any problems with it in 25+ years of intensive programmation. So, they with statement has at least one aficionados: me. Was the with statement present if the first versions designed by Wirth himself ? Cheers, Ched' Le 14. 09. 14 12:20, Marius a écrit : Jürgen Hestermann wrote: Realy? I love it. Consider these two varianst of code from one of my programs: This is a simpler example than the one with with the 4 levels deep (which can be quite a puzzle if there are multiple records containing the same fieldname(s)). And yes you may love them but please read on.. Which one would you prefer? The second, but not for readability or typing, I would suggest a third solution by using a variable (like Matthias also showed). The point is you are very likely to run into refactoring problems. Suppose your routines are in 2 different methods in a class and there is class inheritance and one of the base class has a property errorstring. Suppose you decide to refactor the errorstring in you record structure to something else (and forget to adjust it properly which is human). Your first solution will compile but it would reset the wrong errorstring, your second solution will simply not compile because it has been prefixed (hence my preference to pick that one). You can argue its the fault of the programmer but why create an unsafe situation in the first place (and that just what the with statement does)... We have spend weeks (if not months) off hunting down this kind of bugs. And its this kind of bug what showed up frequently in teams of programmers and that made us decide to ban the with statement as much as possible. Sooner or later the with stament is going to bite you, I just hope you wont have to spend that much time on resolving it ;( Marius ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Ched wrote: Hello All, My two cents. The with statement is crystal clear for me. If newbies prefer to use different manners to code the things, that are free to do it. Using some kind of intermediate variable ? That's make to code difficult to understand, need do open begin/end blocks in some situations. With make the code very clear, probably helps the compiler a lot. Well used, it's 100% safe - I *never* encountered any problems with it in 25+ years of intensive programmation. The consensus is that it has to be used with caution. You get one vote, less a bit for calling those that disagree with you newbies :-) I tend to use it if it will save more than around three repeated prefixes, but I'd be far happier if there were provision for declaring a temporary shortcut symbol: with shortcut: TSomething= TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]^ do begin shortcut.DirLogged := true; shortcut.DirHatFocus := false; shortcut.SubDirsExpanded := true; .. So, they with statement has at least one aficionados: me. Was the with statement present if the first versions designed by Wirth himself ? I believe so, and he also overlooked the dangling else problem. There's also this usage, which has plenty of pitfalls for the unwary or busy: with TSomething.Create do begin .. Free end; -- Mark Morgan Lloyd markMLl .AT. telemetry.co .DOT. uk [Opinions above are the author's, not those of his employers or colleagues] ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
In our previous episode, Ched said: So, they with statement has at least one aficionados: me. Was the with statement present if the first versions designed by Wirth himself ? Yes, but the problems are mostly because of the scoping complexity that OO extensions like classes add. Specially because it is almost custom to use a lot of the same named properties in classes. Something that Wirth couldn't (and didn't have to) forsee. In the past I had the same opinion about with (WITHs evilness is exaggerated) as many others, but once I bumped my head a few times while using OO frameworks that were either not of my design or very large, I basically stopped using it. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Am 14.09.2014 15:39, schrieb Ched: Hello All, My two cents. The with statement is crystal clear for me. If newbies prefer to use different manners to code the things, that are free to do it. Using some kind of intermediate variable ? That's make to code difficult to understand, need do open begin/end blocks in some situations. With make the code very clear, probably helps the compiler a lot. Well used, it's 100% safe - I *never* encountered any problems with it in 25+ years of intensive programmation. So, they with statement has at least one aficionados: me. Was the with statement present if the first versions designed by Wirth himself ? WITH was present in the first PASCAL versions, and it was used in the PASCAL compilers designed by Wirth et al. written in PASCAL in the late 60s and early 70s. The possible dangers of using WITH are only very limited, compared to other languages, for example PL/1, where - you don't need to declare variables, like in FORTRAN - you can omit structure qualifiers, if the the structure components alone are unique - if the structure component is A.B.C, you can write A.C, B.C or simply C, if it is unique in the program - if there is both a simple variable C and a structure component A.C (C in an outer block), I guess, A.C will be used, if you write only C (but I am not sure on this; hopefully there will be a compiler warning) - if you have a combination of vectors and structures, the order of structure components and vector indices is not fixed, that is: A.C(N,M) is the same as A(N).C(M) is the same as A(N,M).C ... regardless of the definition of the structures / vectors A and C. You may read some old articles from Dijkstra and C.A.R. Hoare on this topic, covering PL/1 ... it is very hard to write reliable programs using this language, but it is still in use until today in very large banks and insurance companies. I am using it in my everyday work. So: I would use the PASCAL WITH statement with care, knowing its problems. PASCAL is far less dangerous than PL/1. The other language which I use much is C. There you have other problems that PASCAL (and PL/1) don't have, that is: vector indices out of range, leading to storage overwrites, and storage leaks, due to heavy use of dynamic memory allocation ... which needs careful testing. Both PASCAL and PL/1 can be used with options that control the vector indices, which is IMHO a great help in the early stages of program development. Kind regards Bernd Cheers, Ched' ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Hi, once I came to troubles with with: procedure TForm1.StringGrid1SelectEditor(Sender: TObject; aCol, aRow: Integer; var Editor: TWinControl); begin with StringGrid1 do begin ... Editor:= ...; ... end; end; because Editor is also public property of TStringGrid. Here would be solution to rename parameter to AEditor. Vojtěch ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
The reason why I made these comparisons to PL/1: PL/1 was designed in the mid 60s, and Wirth of course knew about PL/1, because he was at IBM at the time, and of course he was aware of the problems and pitfalls of PL/1. By including the WITH statement into PASCAL, he probably wanted to give the coders some possibilities for an abbreviation (similar to PL/1, but avoiding most of the problems). As we look at it today, not all problems have been avoided. Of course, most inspirations for PASCAL came from ALGOL 60. But ALGOL 60 had no structures, so PL/1 was the natural source of inspiration, at that time. Kind regards Bernd Am 14.09.2014 16:30, schrieb Bernd Oppolzer: The possible dangers of using WITH are only very limited, compared to other languages, for example PL/1, where - you don't need to declare variables, like in FORTRAN - you can omit structure qualifiers, if the the structure components alone are unique - if the structure component is A.B.C, you can write A.C, B.C or simply C, if it is unique in the program - if there is both a simple variable C and a structure component A.C (C in an outer block), I guess, A.C will be used, if you write only C (but I am not sure on this; hopefully there will be a compiler warning) - if you have a combination of vectors and structures, the order of structure components and vector indices is not fixed, that is: A.C(N,M) is the same as A(N).C(M) is the same as A(N,M).C ... regardless of the definition of the structures / vectors A and C. You may read some old articles from Dijkstra and C.A.R. Hoare on this topic, covering PL/1 ... it is very hard to write reliable programs using this language, but it is still in use until today in very large banks and insurance companies. I am using it in my everyday work. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On Sunday 14 September 2014 16:34:06 Marco van de Voort wrote: In the past I had the same opinion about with (WITHs evilness is exaggerated) as many others, but once I bumped my head a few times while using OO frameworks that were either not of my design or very large, I basically stopped using it. Why does Free Pascal not invent a safe with similar as the with of MSElang? If I would be cynical I could write because one must wait until Delphi has it?. ;-) Martin ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
In our previous episode, Martin Schreiber said: basically stopped using it. Why does Free Pascal not invent a safe with similar as the with of MSElang? The primary question is of course if there is a need for a new construct at all. I don't know MSELang, but many new languages try to add new baroque constructs in a desperate attempt to draw users. I always found that doubtful. If I would be cynical I could write because one must wait until Delphi has it?. ;-) Well, then if I were cynical, I would conclude that you don't read compatibility discussions carefully enough. A shame for somebody who wants to implement languages. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
main problem seems to be using property whose name is common within 2 objects of the same with (or code within several with) could it be controlled by a compiler option? in that case compiler may return a hint or warning or even an error depending of the choosen configuration ... does this turn it safe, or safer? Philippe Em 14.09.2014 12:06, Martin Schreiber escreveu: On Sunday 14 September 2014 16:34:06 Marco van de Voort wrote: In the past I had the same opinion about with (WITHs evilness is exaggerated) as many others, but once I bumped my head a few times while using OO frameworks that were either not of my design or very large, I basically stopped using it. Why does Free Pascal not invent a safe with similar as the with of MSElang? If I would be cynical I could write because one must wait until Delphi has it?. ;-) Martin ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [1] Links: -- [1] http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
El 14/09/14 17:20, Marco van de Voort ha escrit: In our previous episode, Martin Schreiber said: basically stopped using it. Why does Free Pascal not invent a safe with similar as the with of MSElang? The primary question is of course if there is a need for a new construct at all. I don't know MSELang, but many new languages try to add new baroque constructs in a desperate attempt to draw users. I always found that doubtful. I replied this morning about TI pascal (ca.1979), but for some reason I don't see it in the list. Anyway https://archive.org/stream/bitsavers_titm990100_41473222/MP351_MPP_UsersManual#page/n183/mode/2up Bye -- Luca ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
someone wrote about a better performance using with. is that true? even with a simple pointer as in: with ptr^ do begin prop1 := ... prop2 := ... end; which should be faster then ptr^.prop1 := ... ptr^.prop1 := ... others wrote it is just usefull to save writing time ... Em 14.09.2014 12:40, Luca Olivetti escreveu: El 14/09/14 17:20, Marco van de Voort ha escrit: In our previous episode, Martin Schreiber said: basically stopped using it. Why does Free Pascal not invent a safe with similar as the with of MSElang? The primary question is of course if there is a need for a new construct at all. I don't know MSELang, but many new languages try to add new baroque constructs in a desperate attempt to draw users. I always found that doubtful. I replied this morning about TI pascal (ca.1979), but for some reason I don't see it in the list. Anyway https://archive.org/stream/bitsavers_titm990100_41473222/MP351_MPP_UsersManual#page/n183/mode/2up [1] Bye Links: -- [1] https://archive.org/stream/bitsavers_titm990100_41473222/MP351_MPP_UsersManual#page/n183/mode/2up ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
El 14/09/14 12:22, Jürgen Hestermann ha escrit: - d:=TreeRoot.SubDirs[Low(TreeRoot.SubDirs)]; // local var d d^.DirLogged := true; d^.DirHatFocus := false; d^.SubDirsExpanded := true; [...] - Long-winded and still not the same clarity as in the with statement. As I say every time this discussion comes up ;-), TI pascal had the best of both worlds: you could either use the naked with or a named with, which removed ambiguity, i.e. with a=Label1,b=Label2 do a.caption:=b.caption; (OK, it's a contrived example but you get the drift). Bye -- Luca ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 9/13/2014 2:02 PM, Sven Barth wrote: It has always worked in this manner. It's just for abbreviating typing, not for combining properties. (and then the with-variant with multiple elements is even more seldomly used than the single-element one...) In the Turbo Pascal days, WITH allowed for some user-directed compiler optimization. From the TP7 manual: In certain situations, Turbo Pascal's code generator can eliminate redundant pointer-load instructions, shrinking the size of the code and allowing for faster execution. When the code generator can guarantee that a particular pointer remains constant over a stretch of linear code (code with no jumps into it), and when that pointer is already loaded into a register pair (such as ES:DI), the code generator eliminates additional redundant pointer-load instructions in that block of code. A pointer is considered constant if it's obtained from a variable parameter (variable parameters are always passed as pointers) or from the variable reference of a WITH statement. Because of this, using WITH statements is often more efficient (but never less efficient) than writing the fully-qualified variable for each component reference. -- Jim Leonard (trix...@oldskool.org) Check out some trippy MindCandy: http://www.mindcandydvd.com/ A child borne of the home computer wars: http://trixter.oldskool.org/ You're all insane and trying to steal my magic bag! ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 14.09.2014 18:05, Philippe wrote: someone wrote about a better performance using with. is that true? even with a simple pointer as in: with ptr^ do begin prop1 := ... prop2 := ... end; which should be faster then ptr^.prop1 := ... ptr^.prop1 := ... others wrote it is just usefull to save writing time ... Take this example: === code begin === program twithtest; type TTest = object procedure SetProp1(aValue: Integer); procedure SetProp2(aValue: Integer); property Prop1: Integer write SetProp1; property Prop2: Integer write SetProp2; end; PTest = ^TTest; procedure TTest.SetProp1(aValue: Integer); begin end; procedure TTest.SetProp2(aValue: Integer); begin end; procedure TestWith; var p: PTest; begin New(p); with p^ do begin Prop1 := 42; Prop2 := 21; end; Dispose(p); end; procedure TestWithout; var p: PTest; begin New(p); p^.Prop1 := 42; p^.Prop2 := 21; Dispose(p); end; begin end. === code end === This is the relevant code generated for TestWith: === asm begin === # [28] with p^ do begin movl-4(%ebp),%ebx # [29] Prop1 := 42; movl%ebx,%eax movw$42,%dx callP$TWITHTEST_TTEST_$__SETPROP1$SMALLINT # [30] Prop2 := 21; movl%ebx,%eax movw$21,%dx callP$TWITHTEST_TTEST_$__SETPROP2$SMALLINT === asm end === and this for TestWithout: === asm begin === # [42] p^.Prop1 := 42; movl-4(%ebp),%eax movw$42,%dx callP$TWITHTEST_TTEST_$__SETPROP1$SMALLINT # [43] p^.Prop2 := 21; movl-4(%ebp),%eax movw$21,%dx callP$TWITHTEST_TTEST_$__SETPROP2$SMALLINT === asm end === As you can see the expression p^ is only evaluated once in the TestWith case while it's evaluated twice in the TestWithout one. So it's only minimally faster in this example (one less memory access), but if you use enough members of TTest it a more or less tight loop it might even be noticeable. Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 15.09.2014 03:06, Jim Leonard wrote: On 9/13/2014 2:02 PM, Sven Barth wrote: It has always worked in this manner. It's just for abbreviating typing, not for combining properties. (and then the with-variant with multiple elements is even more seldomly used than the single-element one...) In the Turbo Pascal days, WITH allowed for some user-directed compiler optimization. From the TP7 manual: In certain situations, Turbo Pascal's code generator can eliminate redundant pointer-load instructions, shrinking the size of the code and allowing for faster execution. When the code generator can guarantee that a particular pointer remains constant over a stretch of linear code (code with no jumps into it), and when that pointer is already loaded into a register pair (such as ES:DI), the code generator eliminates additional redundant pointer-load instructions in that block of code. A pointer is considered constant if it's obtained from a variable parameter (variable parameters are always passed as pointers) or from the variable reference of a WITH statement. Because of this, using WITH statements is often more efficient (but never less efficient) than writing the fully-qualified variable for each component reference. FPC optimizes this as well (because the with-expression is only evaluated once). Though it's arguable in how far that is a micro optimization with today's CPUs. Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
[fpc-pascal] with statement using mulltiple objects
According to the docs the with statement when used with mulltiple objects only works with the last parameter. The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; -- I thought that if all the objects have the property the statement would apply to them eg. with Label1, Label2, Label3 do Caption = 'Text'. I thought all the 3 labels would have the caption set, but only Label3's caption is set. Is this a difference between Delphi's Object Pascal and FreePascal, or has Delphi's Object Pascal always worked in the same manner? Frank Church === http://devblog.brahmancreations.com ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 13.09.2014 20:13, vfclists . wrote: According to the docs the with statement when used with mulltiple objects only works with the last parameter. The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; -- I thought that if all the objects have the property the statement would apply to them eg. with Label1, Label2, Label3 do Caption = 'Text'. I thought all the 3 labels would have the caption set, but only Label3's caption is set. Is this a difference between Delphi's Object Pascal and FreePascal, or has Delphi's Object Pascal always worked in the same manner? It has always worked in this manner. It's just for abbreviating typing, not for combining properties. (and then the with-variant with multiple elements is even more seldomly used than the single-element one...) Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 13 September 2014 20:02, Sven Barth pascaldra...@googlemail.com wrote: On 13.09.2014 20:13, vfclists . wrote: According to the docs the with statement when used with mulltiple objects only works with the last parameter. The statement With A,B,C,D do Statement; is equivalent to With A do With B do With C do With D do Statement; -- I thought that if all the objects have the property the statement would apply to them eg. with Label1, Label2, Label3 do Caption = 'Text'. I thought all the 3 labels would have the caption set, but only Label3's caption is set. Is this a difference between Delphi's Object Pascal and FreePascal, or has Delphi's Object Pascal always worked in the same manner? It has always worked in this manner. It's just for abbreviating typing, not for combining properties. (and then the with-variant with multiple elements is even more seldomly used than the single-element one...) Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal What then is the nesting for if the command will only apply to the last item with the property? Is each nested element supposed to be a property of the enclosing element? What is the rationale for such a statement, ie using multliple elements? -- Frank Church === http://devblog.brahmancreations.com ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
On 13.09.2014 21:29, vfclists . wrote: What then is the nesting for if the command will only apply to the last item with the property? Is each nested element supposed to be a property of the enclosing element? What is the rationale for such a statement, ie using multliple elements? Consider this example: === code begin === type TTest1 = class procedure Foo; end; TTest2 = class proceduer Bar; end; // implementation of TTest1 and TTest2 var t1: TTest1; t2: TTest2; begin with t1, t2 do begin Foo; // calls t1.Foo Bar; // calls t2.Bar end; // same as: with t1 do begin with t2 do begin Foo; Bar; end; end; end. === code end === with simply allows you to avoid some typing and evaluatios the expression between the with and do only once (take a function for example instead of t1). So mostly it's merely syntactic sugar. Regards, Sven ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
Mattias Gaertner wrote: Less typing. Just some generic comments in general. Its also much harder to understand in the long run and it can even be dangerous as stuff migh be refactored over time. All seasoned programmers know that and try to avoid the with statement. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] with statement using mulltiple objects
El 13/09/14 20:13, vfclists . ha escrit: I thought all the 3 labels would have the caption set, but only Label3's caption is set. Is this a difference between Delphi's Object Pascal and FreePascal, or has Delphi's Object Pascal always worked in the same manner? The with syntax comes from plain old pascal and it has always worked this way. http://books.google.co.uk/books?id=xXSZbSLFTM8Clpg=PP1hl=capg=PA73#v=onepageqf=false Bye -- Luca ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
