grahamd@otc.otca.oz.au (Graham Dumpleton) (04/23/91)
I checked the FAQ list but couldn't find this mentioned so: What is an ANSI conforming C preprocessor mean't to do in the following situation. #define preone onepre #define one two #define addpre(arg) pre##arg addpre(one) The preprocessor on a RS6000 running AIX when run with the argument '-qlanglvl=ansi' to invoke ANSI compatability produces onepre In other words it performs the join and then checks to see if the result is a macro and if so expands it. The pre-processor which we use though, which is based on the Decus cpp supplied on the X tape produces the following pretwo In other words it has the effect of expanding the argument if it is a macro first and then performing the join. I tend to think that the Decus cpp is wrong but can someone please confirm this. -- Graham Dumpleton (grahamd@otc.otca.oz.au)
steve@taumet.com (Stephen Clamage) (04/23/91)
grahamd@otc.otca.oz.au (Graham Dumpleton) writes: > #define preone onepre > #define one two > #define addpre(arg) pre##arg > addpre(one) >The preprocessor on a RS6000 running AIX when run with the argument >'-qlanglvl=ansi' to invoke ANSI compatability produces > onepre >The pre-processor which we use though, which is based on the Decus cpp supplied >on the X tape produces the following > pretwo >I tend to think that the Decus cpp is wrong but can someone please confirm this. Decus is wrong. Section 3.8.3.1 of the Standard decribes how macro argument substitution operates. After identifying the arguments, those which are neither preceeded by # or ##, nor followed by ##, are fully macro-replaced; the others are left alone. In your example, the 'arg' macro is preceeded by ##, so its actual parameter is not expanded. After concatenation, we have 'preone', which is then rescanned to produce 'onepre'. -- Steve Clamage, TauMetric Corp, steve@taumet.com
sarima@tdatirv.UUCP (Stanley Friesen) (04/24/91)
In article <2610@otc.otca.oz> grahamd@otc.otca.oz.au (Graham Dumpleton) writes:
<What is an ANSI conforming C preprocessor mean't to do in the following
<situation.
<
<#define preone onepre
<#define one two
<#define addpre(arg) pre##arg
<
<addpre(one)
I believe that the ANSI standard requires that macro arguments be expanded
before substitution into the macro body, and therefor before expansion of
the macro itself into the source.
Thus the answer is:
pretwo
<The preprocessor on a RS6000 running AIX ...
<onepre
I believe that this is *wrong* per the standard.
(I do not actually have a copy present with me, so I could be wrong).
--
---------------
uunet!tdatirv!sarima (Stanley Friesen)sparks@suntan.viewlogic.com (Alan Sparks) (05/04/91)
In article <208@tdatirv.UUCP>, sarima@tdatirv.UUCP (Stanley Friesen) writes: |> In article <2610@otc.otca.oz> grahamd@otc.otca.oz.au (Graham Dumpleton) writes: |> <What is an ANSI conforming C preprocessor mean't to do in the following |> <situation. |> < |> <#define preone onepre |> <#define one two |> <#define addpre(arg) pre##arg |> < |> <addpre(one) |> |> I believe that the ANSI standard requires that macro arguments be expanded |> before substitution into the macro body, and therefor before expansion of |> the macro itself into the source. |> |> Thus the answer is: |> |> pretwo |> |> <The preprocessor on a RS6000 running AIX ... |> |> <onepre |> |> I believe that this is *wrong* per the standard. |> (I do not actually have a copy present with me, so I could be wrong). |> -- |> --------------- |> uunet!tdatirv!sarima (Stanley Friesen) Paragraphs in the Standard suggest that AIX is correct (section A12.3): "... Unless the parameter in the replacement sequence is preceded by #, or preceded or followed by ##, the argument tokens are examined for macro calls and expanded as necessary, just before insertion." ".. Second, if the definition token sequence ... contains a ## operator, then just after replacement of the parameters, each ## is deleted, together with any whitespace on either side, so as to concatenate the adjacent tokens and form a new token." "In both kinds of macro, the replacement token sequence is repeatedly rescanned for more defined identifiers." So, this suggests that addpre(one) turns into pre ## one and (after concatenation), preone and finally, after rescan, onepre -Alan -- Alan Sparks voice: (508) 480-0881 Internet: sparks@viewlogic.com VIEWlogic Systems Inc., 293 Boston Post Rd. W., Marlboro MA 01752 Disclaimer: VIEWlogic didn't say this; I might have. Who's asking?