Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.lang.clipper.visual-objects > #5283 > unrolled thread
| Started by | kvstegemann@web.de |
|---|---|
| First post | 2013-05-27 08:53 -0700 |
| Last post | 2013-05-31 03:43 -0700 |
| Articles | 4 on this page of 24 — 8 participants |
Back to article view | Back to comp.lang.clipper.visual-objects
Shortint calculations kvstegemann@web.de - 2013-05-27 08:53 -0700
Re: Shortint calculations Karl Faller <mail_withoutthat_@_drei-f_andthat_.de> - 2013-05-27 18:09 +0200
Re: Shortint calculations dlzc <dlzc1@cox.net> - 2013-05-27 10:21 -0700
Re: Shortint calculations Karl Faller <mail_withoutthat_@_drei-f_andthat_.de> - 2013-05-27 20:32 +0200
Re: Shortint calculations kvstegemann@web.de - 2013-05-27 11:46 -0700
Re: Shortint calculations Karl Faller <mail_withoutthat_@_drei-f_andthat_.de> - 2013-05-27 23:40 +0200
Re: Shortint calculations "Wolfgang Riedmann" <wriedmann@gmail.com> - 2013-05-28 08:02 +0200
Re: Shortint calculations Mathias <mathias.hakansson@consultec.se> - 2013-05-27 22:39 -0700
Re: Shortint calculations Karl Faller <mail_withoutthat_@_drei-f_andthat_.de> - 2013-05-28 09:01 +0200
Re: Shortint calculations kvstegemann@web.de - 2013-05-28 01:02 -0700
Re: Shortint calculations Karl Faller <mail_withoutthat_@_drei-f_andthat_.de> - 2013-05-28 10:32 +0200
Re: Shortint calculations kvstegemann@web.de - 2013-05-28 01:46 -0700
Re: Shortint calculations Meinhard Schnoor-Matriciani <msm@appfact.de> - 2013-05-28 11:45 +0200
Re: Shortint calculations Mathias <mathias.hakansson@consultec.se> - 2013-05-28 23:45 -0700
Re: Shortint calculations "Er!k \\/!$$Er" <nospam@nospam.url> - 2013-05-29 16:57 +0200
Re: Shortint calculations dlzc <dlzc1@cox.net> - 2013-05-29 09:33 -0700
Re: Shortint calculations dlzc <dlzc1@cox.net> - 2013-05-28 07:15 -0700
Re: Shortint calculations Mathias <mathias.hakansson@consultec.se> - 2013-05-29 04:44 -0700
Re: Shortint calculations dlzc <dlzc1@cox.net> - 2013-05-29 07:25 -0700
Re: Shortint calculations "Jamal" <jamal@examples.net> - 2013-05-29 17:18 -0400
Re: Shortint calculations kvstegemann@web.de - 2013-05-30 01:47 -0700
Re: Shortint calculations "Jamal" <jamal@examples.net> - 2013-05-30 13:19 -0400
Re: Shortint calculations Meinhard Schnoor-Matriciani <msm@appfact.de> - 2013-05-30 22:05 +0200
Re: Shortint calculations kvstegemann@web.de - 2013-05-31 03:43 -0700
Page 2 of 2 — ← Prev page 1 [2]
| From | kvstegemann@web.de |
|---|---|
| Date | 2013-05-30 01:47 -0700 |
| Message-ID | <a168c874-1b2f-4438-93f2-33ed3dbfa607@googlegroups.com> |
| In reply to | #5305 |
Am Mittwoch, 29. Mai 2013 23:18:10 UTC+2 schrieb Jamal: > So try putting the complex expression part in left hand of the comparison. Jamal, there's really no more need of another workaround. If I want a workaround, I just use longints. The question is, where to use workarounds? What exactly is broken? Is it the > operator with shortints? Is it the evaluation of simple operations like subtractions with shortints? (Not really a "complex expression" ...) Is it an exotic combination of expressions? I hope Meinhard finds out, so that we can spot the places where a workaround is needed. And then of course, a fix would be nice. Have a nice day Kay-Viktor
[toc] | [prev] | [next] | [standalone]
| From | "Jamal" <jamal@examples.net> |
|---|---|
| Date | 2013-05-30 13:19 -0400 |
| Message-ID | <ko81m4$7k7$1@speranza.aioe.org> |
| In reply to | #5307 |
Kay-Victor, Good luck waiting for a solution. The compiler people like Fru left the Dev team since CA abandoned the product. Jamal wrote in message news:a168c874-1b2f-4438-93f2-33ed3dbfa607@googlegroups.com... Am Mittwoch, 29. Mai 2013 23:18:10 UTC+2 schrieb Jamal: > So try putting the complex expression part in left hand of the comparison. Jamal, there's really no more need of another workaround. If I want a workaround, I just use longints. The question is, where to use workarounds? What exactly is broken? Is it the > operator with shortints? Is it the evaluation of simple operations like subtractions with shortints? (Not really a "complex expression" ...) Is it an exotic combination of expressions? I hope Meinhard finds out, so that we can spot the places where a workaround is needed. And then of course, a fix would be nice. Have a nice day Kay-Viktor
[toc] | [prev] | [next] | [standalone]
| From | Meinhard Schnoor-Matriciani <msm@appfact.de> |
|---|---|
| Date | 2013-05-30 22:05 +0200 |
| Message-ID | <51a7b3da$0$6626$3ca0ca56@newsspool2.vodafone-ip.de> |
| In reply to | #5283 |
Hi Kay-Victor,
just looked into the compiler to see what happens here. Roughly spoken
it happens the following (it is quite tricky to find these things out as
there is nearly no documentation) :
- the arithmetic expression is being evaluated on 32bit base even
shortints are only 16bits
- the comparison is done by using the right hand intermediate datatype
which is dword in this case ( which is wrong I admit, internally the
32bit registers are used and the values in there are just dwords ) I you
example a value of 0xfffff3 is used for the comparison and this is the
reason while the comparison fails
If you turn it the other way round ( nB - nC ) <= nA it works because
the result of (nB-nC) get converted to SHORTINT before the comparison
takes place because the right hand part results in the datatype SHORINT.
Another, abstruse way, to get around the problem is to use
IF nA > SHORTINT(DWORD((nB - nC ))) ....
I have checked the old versions as well and this problem is there at
least since 2.0 ....
Regards
Meinhard
Am 27.05.2013 17:53, schrieb kvstegemann@web.de:
> FUNCTION Start()
>
> LOCAL nA, nB, nC AS SHORTINT
>
> nA := 748
> nB := 22
> nC := 35
>
> IF nA > ( nB - nC )
> ? "748 is greater than (22 - 35)"
> ELSE
> ? "748 is not greater than (22 - 35). Oops!"
> ENDIF
>
> _Wait("")
--
Meinhard Schnoor-Matriciani
Grafx Development Team
[toc] | [prev] | [next] | [standalone]
| From | kvstegemann@web.de |
|---|---|
| Date | 2013-05-31 03:43 -0700 |
| Message-ID | <9fc4c357-04af-4e48-9abf-c75c5b63ea94@googlegroups.com> |
| In reply to | #5311 |
Hi Meinhard, thanks for checking that. So when I want to play it safe, I could say as a rule of thumb, "don't use expressions containing shortints on the right side of a comparison operator". Right? In other words, I could prevent this sort of problem by either - not using shortints at all, or - evaluating shortint expressions into a separate variable before applying a comparison operator (The "nD" approach :) Have a nice weekend Kay-Viktor
[toc] | [prev] | [standalone]
Page 2 of 2 — ← Prev page 1 [2]
Back to top | Article view | comp.lang.clipper.visual-objects
csiph-web