[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    [PATCH] Changes in Strigi caused build problems in 4.1 BRANCH
From:       James Richard Tyrer <tyrerj () acm ! org>
Date:       2008-08-26 13:01:58
Message-ID: 48B3FEC6.1030708 () acm ! org
[Download RAW message or body]

James Richard Tyrer wrote:
> This is from KDELibs:
> 
> [ 45%] Building CXX object 
> kio/kio/dummyanalyzers/CMakeFiles/dummy.dir/dummyanalyzers.o
> /usr/SVN/branches/KDE/4.1/kdelibs/kio/kio/dummyanalyzers/dummyanalyzers.cpp:43: 
> error: conflicting return type specified for 'virtual char 
> DummyEndAnalyzer::analyze(Strigi::AnalysisResult&, Strigi::InputStream*)'
> /usr/local/KDE-4.1/include/strigi/streamendanalyzer.h:37: error: 
> overriding 'virtual signed char 
> Strigi::StreamEndAnalyzer::analyze(Strigi::AnalysisResult&, 
> Strigi::InputStream*)'
> make[2]: *** 
> [kio/kio/dummyanalyzers/CMakeFiles/dummy.dir/dummyanalyzers.o] Error 1
> make[1]: *** [kio/kio/dummyanalyzers/CMakeFiles/dummy.dir/all] Error 2
> make: *** [all] Error 2
> 
It should seem obvious but the types do need to match, and you can't 
assign the value "-1" to unsigned character.

------8<------8<------8<------8<------8<------8<------8<------8<------

Index: kio/kio/dummyanalyzers/dummyanalyzers.cpp
===================================================================
--- kio/kio/dummyanalyzers/dummyanalyzers.cpp   (revision 852451)
+++ kio/kio/dummyanalyzers/dummyanalyzers.cpp   (working copy)
@@ -40,7 +40,7 @@ public:
      bool checkHeader(const char*, int32_t) const {
          return false;
      }
-    char analyze(Strigi::AnalysisResult&, InputStream*) {
+    signed char analyze(Strigi::AnalysisResult&, InputStream*) {
          return -1;
      }
      const char* name() const { return "DummyEndAnalyzer"; }


------8<------8<------8<------8<------8<------8<------8<------8<------

Index: strigi-analyzer/rfc822/rfc822endanalyzer.cpp
===================================================================
--- strigi-analyzer/rfc822/rfc822endanalyzer.cpp        (revision 852459)
+++ strigi-analyzer/rfc822/rfc822endanalyzer.cpp        (working copy)
@@ -38,7 +38,7 @@ bool Rfc822EndAnalyzer::checkHeader(  co
      return false;
  }

-char Rfc822EndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
+signed char Rfc822EndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
  {
      char id_from[] = "From: ";
      char id_to[] = "To: ";
Index: strigi-analyzer/rfc822/rfc822endanalyzer.h
===================================================================
--- strigi-analyzer/rfc822/rfc822endanalyzer.h  (revision 852459)
+++ strigi-analyzer/rfc822/rfc822endanalyzer.h  (working copy)
@@ -36,7 +36,7 @@ public:

    const char* name() const { return "Rfc822EndAnalyzer"; }
    bool checkHeader( const char* header, int32_t headersize ) const;
-  char analyze(  Strigi::AnalysisResult& idx, Strigi::InputStream* in );
+  signed char analyze(  Strigi::AnalysisResult& idx, 
Strigi::InputStream* in );

  private:
    const Rfc822EndAnalyzerFactory* m_factory;
Index: strigi-analyzer/ics/icsendanalyzer.h
===================================================================
--- strigi-analyzer/ics/icsendanalyzer.h        (revision 852459)
+++ strigi-analyzer/ics/icsendanalyzer.h        (working copy)
@@ -36,7 +36,7 @@ public:

    const char* name() const { return "IcsEndAnalyzer"; }
    bool checkHeader( const char* header, int32_t headersize ) const;
-  char analyze(  Strigi::AnalysisResult& idx, Strigi::InputStream* in );
+  signed char analyze(  Strigi::AnalysisResult& idx, 
Strigi::InputStream* in );

  private:
    const IcsEndAnalyzerFactory* m_factory;
Index: strigi-analyzer/ics/icsendanalyzer.cpp
===================================================================
--- strigi-analyzer/ics/icsendanalyzer.cpp      (revision 852459)
+++ strigi-analyzer/ics/icsendanalyzer.cpp      (working copy)
@@ -51,7 +51,7 @@ bool IcsEndAnalyzer::checkHeader( const
  I chose to use libkcal instead of reading the calendar manually. It's 
easier to
  maintain this way.
  */
-char IcsEndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
+signed char IcsEndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
  {
    CalendarLocal cal( QString::fromLatin1( "UTC" ) );

Index: strigi-analyzer/vcf/vcfendanalyzer.h
===================================================================
--- strigi-analyzer/vcf/vcfendanalyzer.h        (revision 852459)
+++ strigi-analyzer/vcf/vcfendanalyzer.h        (working copy)
@@ -37,7 +37,7 @@ class PIMSTRIGI_ANALYZER_EXPORT VcfEndAn

      const char* name() const { return "VcfEndAnalyzer"; }
      bool checkHeader( const char* header, int32_t headersize ) const;
-    char analyze(  Strigi::AnalysisResult& idx, Strigi::InputStream* in );
+    signed char analyze(  Strigi::AnalysisResult& idx, 
Strigi::InputStream* in );

    private:
      const VcfEndAnalyzerFactory* m_factory;
Index: strigi-analyzer/vcf/vcfendanalyzer.cpp
===================================================================
--- strigi-analyzer/vcf/vcfendanalyzer.cpp      (revision 852459)
+++ strigi-analyzer/vcf/vcfendanalyzer.cpp      (working copy)
@@ -43,7 +43,7 @@ bool VcfEndAnalyzer::checkHeader( const
   * It's easier to use KABC::VCardConverter to extract the single 
fields from the vCard
   * than doing it manually.
   */
-char VcfEndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
+signed char VcfEndAnalyzer::analyze( Strigi::AnalysisResult& idx, 
Strigi::InputStream* in )
  {
    const char* data;
    if ( in->read( data, 1, in->size() ) < 0 )

------8<------8<------8<------8<------8<------8<------8<------8<------

Although I don't understand the code -- I only fixed the conflicting 
types -- and don't know if it is wise to use the signed character type 
here, this does fix the compilation problems.  If it causes other bugs, 
then the code needs some work.  Unless we are actually dealing with 7 
bit signed integer data, I would always suggest using an unsigned 
character and a Boolean variable.  Such code is simply a better way to 
do things (easier to understand and maintain, etc.).  If we must have a 
single return value, use a structure.

-- 
JRT
 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
[prev in list] [next in list] [prev in thread] [next in thread]