You are viewing a plain text version of this content. The canonical link for it is here.
Posted to c-dev@axis.apache.org by John Hawkins <HA...@uk.ibm.com> on 2004/05/27 14:54:29 UTC
Build issues
Hi Folks,
we seem to be getting into some problems with the build at the moment.
Build stability is an issue if we are to direct people to just take the
latest drops. Do you have a regular automated build process? It doesn't
have to have tests attached just something that merely builds say, on the
hour every hour and has a log is good enough to verify that we can be happy
to direct folks to the CVS source. We have a number of tools that can help
you if you need them that allow you to extract from CVS to a machine and
build on a regular basis.
What do you think? It breaks my little 'ol heart to see people using gSoap
!
John
Re: Build issues
Posted by Kenneth Chiu <ch...@cs.indiana.edu>.
On Mon, 31 May 2004 damitha@opensource.lk wrote:
> Hi,
>
> > John Hawkins wrote:
> >
> >>we seem to be getting into some problems with the build at the moment.
> >>Build stability is an issue if we are to direct people to just take the
> >>latest drops. Do you have a regular automated build process? It doesn't
> >>have to have tests attached just something that merely builds say, on the
> >>hour every hour and has a log is good enough to verify that we can be
> >> happy
> >>to direct folks to the CVS source. We have a number of tools that can
> >> help
> >>you if you need them that allow you to extract from CVS to a machine and
> >>build on a regular basis.
> >>
> >>What do you think? It breaks my little 'ol heart to see people using
> >> gSoap
> >>!
> >>
> >>
> > agreed. this process is way way too complicated!!!!!!!!!!!!!!!
> >
> > here are my impressions after wasting 6+ hours on that and still failed!
> >
> > maybe automatic builds should be REQUIRED but who can automate builds
> > out of CVS if there is so many manual steps!
> >
> > here is what are my impression - i have fresh version checked out from
> > CVS and i have not expected it will be so complicated to build it ...
> >
> > something like 20+ steps (feels like hundreds of steps ...) to configure
> > AXIS-C++? this is insaneley complicated ..
>
> Samisa has written a script file which build and test samples in one shot.
> But that is for automated building and testing. It is not suitable for
> building and testing while you develop. Yes we need to look more into
> making the build process more easier. I feel that it is not a difficult
> thing to make build process very easy and simple. Only somebody need to
> attend to it.
>
> >
> > why on earth there is a separate autogen/runconfig for ws-axis/c,
> > samples/client, and samples/server ?!
>
> OK, now you have only one build.sh file to build all stuff.
>
> >
> > why configure script is not checked into CVS?! isnt it whole purpose of
> > configure script to ease that process?
>
> I thought that it is the old way of doing building. Since we write only
> configure.ac and configure is generated by auto tools I thought that it is
> not neccesary to put it in cvs.
The problem though is that then this requires people who
compile out of CVS to have the right autotools, etc.
We have found that compiling out of CVS requires us to learn
a lot more about autotools than we really want to know.
> If we are not to use GNU auto tools then we need to write a configure file
> which tackle many platforms and put it into CVS.
>
> > why not have configure to correclty set apache1/2, expat/xercesc/both in
> > ./configure intead of manually hacking
>
> Yes we need to that.
>
> >
> > why there are so many warnings when compiling sampleS?
>
> Most of the code at sampels are generated ones. I think we eed to work on
> wsdl2ws tool to remove them. But for the time being I think we need to
> remove those warnings manually.
>
> > it seems that SOAPTransportFactory::initialize() has hardcoded
> > /usr/local/Axis/ so unless i change it i get this:
>
> Yes this is still hard coded. it is supposed to read from the
> configuration file. But still work is going on.
>
> >
> > thanks,
> >
> > alek
> >
> > --
> > The best way to predict the future is to invent it - Alan Kay
> >
> >
> >
> >
> > --
> > The best way to predict the future is to invent it - Alan Kay
>
>
> thanks
>
> damitha
>
Re: Build issues
Posted by da...@opensource.lk.
Hi,
> John Hawkins wrote:
>
>>we seem to be getting into some problems with the build at the moment.
>>Build stability is an issue if we are to direct people to just take the
>>latest drops. Do you have a regular automated build process? It doesn't
>>have to have tests attached just something that merely builds say, on the
>>hour every hour and has a log is good enough to verify that we can be
>> happy
>>to direct folks to the CVS source. We have a number of tools that can
>> help
>>you if you need them that allow you to extract from CVS to a machine and
>>build on a regular basis.
>>
>>What do you think? It breaks my little 'ol heart to see people using
>> gSoap
>>!
>>
>>
> agreed. this process is way way too complicated!!!!!!!!!!!!!!!
>
> here are my impressions after wasting 6+ hours on that and still failed!
>
> maybe automatic builds should be REQUIRED but who can automate builds
> out of CVS if there is so many manual steps!
>
> here is what are my impression - i have fresh version checked out from
> CVS and i have not expected it will be so complicated to build it ...
>
> something like 20+ steps (feels like hundreds of steps ...) to configure
> AXIS-C++? this is insaneley complicated ..
Samisa has written a script file which build and test samples in one shot.
But that is for automated building and testing. It is not suitable for
building and testing while you develop. Yes we need to look more into
making the build process more easier. I feel that it is not a difficult
thing to make build process very easy and simple. Only somebody need to
attend to it.
>
> why on earth there is a separate autogen/runconfig for ws-axis/c,
> samples/client, and samples/server ?!
OK, now you have only one build.sh file to build all stuff.
>
> why configure script is not checked into CVS?! isnt it whole purpose of
> configure script to ease that process?
I thought that it is the old way of doing building. Since we write only
configure.ac and configure is generated by auto tools I thought that it is
not neccesary to put it in cvs.
If we are not to use GNU auto tools then we need to write a configure file
which tackle many platforms and put it into CVS.
>
> why not have configure to correclty set apache1/2, expat/xercesc/both in
> ./configure intead of manually hacking
Yes we need to that.
>
> why there are so many warnings when compiling sampleS?
Most of the code at sampels are generated ones. I think we eed to work on
wsdl2ws tool to remove them. But for the time being I think we need to
remove those warnings manually.
> it seems that SOAPTransportFactory::initialize() has hardcoded
> /usr/local/Axis/ so unless i change it i get this:
Yes this is still hard coded. it is supposed to read from the
configuration file. But still work is going on.
>
> thanks,
>
> alek
>
> --
> The best way to predict the future is to invent it - Alan Kay
>
>
>
>
> --
> The best way to predict the future is to invent it - Alan Kay
thanks
damitha
Proxy support for clients
Posted by Samisa Abeysinghe <sa...@yahoo.com>.
Hi All,
I added proxy tunneling support to client stubs.
This is a by-product of my effort to test Axis C++ client side with external web services. (I
had to go through a proxy) Now people can use Axis C++ clinet from behind a proxy.
However, my testing so far on interop with exteranl services have not given me satisfactory
results. SOAP builders round 1 tests fail with .NET service.
Thanks,
Samisa...
__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/
Re: WSDL2Ws considers Service entry to be mandatory
Posted by Samisa Abeysinghe <sa...@yahoo.com>.
Hi Alek,
I am using the latest CVS checkout.
I will try your patch, thanks for the information.
BTW: Could someone who is familiar with the Java tool apply the patch to CVS? Please....
Thanks,
Samisa...
--- Aleksander Slominski <as...@cs.indiana.edu> wrote:
> Samisa Abeysinghe wrote:
>
> >When I use the tool on the generic SOAPBuilders round 1 wsdl (InteropTest.wsdl), the tool
> fails:
> >
> >Exception in thread "main" org.apache.axis.wsdl.wsdl2ws.WrapperFault: the service does not
> exists
> > at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.getServiceEntry(Unknown Source)
> > at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.perprocess(Unknown Source)
> > at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.genarateWrappers(Unknown Source)
> > at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.main(Unknown Source)
> >
> >Looks like it assumes the service to be there always. Is this assumtion reasonable?
> >
> >BTW: I tried the same wsdl with Axis Java's WSDL2Java tool and it works.
> >
> >
> >
> >
>
> hi Samisa,
>
> i think i had the same problem. what version of AXIS-C++ and AXIS-Java jar do you use?
> if you use the latest AXIS-C++/Java from CVS then you may need to apply the patch i sent to the
> mailing list.
>
> HTH,
>
> alek
>
> --
> The best way to predict the future is to invent it - Alan Kay
>
__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/
Re: WSDL2Ws considers Service entry to be mandatory
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
Samisa Abeysinghe wrote:
>When I use the tool on the generic SOAPBuilders round 1 wsdl (InteropTest.wsdl), the tool fails:
>
>Exception in thread "main" org.apache.axis.wsdl.wsdl2ws.WrapperFault: the service does not exists
> at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.getServiceEntry(Unknown Source)
> at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.perprocess(Unknown Source)
> at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.genarateWrappers(Unknown Source)
> at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.main(Unknown Source)
>
>Looks like it assumes the service to be there always. Is this assumtion reasonable?
>
>BTW: I tried the same wsdl with Axis Java's WSDL2Java tool and it works.
>
>
>
>
hi Samisa,
i think i had the same problem. what version of AXIS-C++ and AXIS-Java jar do you use?
if you use the latest AXIS-C++/Java from CVS then you may need to apply the patch i sent to the mailing list.
HTH,
alek
--
The best way to predict the future is to invent it - Alan Kay
WSDL2Ws considers Service entry to be mandatory
Posted by Samisa Abeysinghe <sa...@yahoo.com>.
When I use the tool on the generic SOAPBuilders round 1 wsdl (InteropTest.wsdl), the tool fails:
Exception in thread "main" org.apache.axis.wsdl.wsdl2ws.WrapperFault: the service does not exists
at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.getServiceEntry(Unknown Source)
at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.perprocess(Unknown Source)
at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.genarateWrappers(Unknown Source)
at org.apache.axis.wsdl.wsdl2ws.WSDL2Ws.main(Unknown Source)
Looks like it assumes the service to be there always. Is this assumtion reasonable?
BTW: I tried the same wsdl with Axis Java's WSDL2Java tool and it works.
Samisa...
__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/
Re: Build issues
Posted by Aleksander Slominski <as...@cs.indiana.edu>.
John Hawkins wrote:
>we seem to be getting into some problems with the build at the moment.
>Build stability is an issue if we are to direct people to just take the
>latest drops. Do you have a regular automated build process? It doesn't
>have to have tests attached just something that merely builds say, on the
>hour every hour and has a log is good enough to verify that we can be happy
>to direct folks to the CVS source. We have a number of tools that can help
>you if you need them that allow you to extract from CVS to a machine and
>build on a regular basis.
>
>What do you think? It breaks my little 'ol heart to see people using gSoap
>!
>
>
agreed. this process is way way too complicated!!!!!!!!!!!!!!!
here are my impressions after wasting 6+ hours on that and still failed!
maybe automatic builds should be REQUIRED but who can automate builds
out of CVS if there is so many manual steps!
here is what are my impression - i have fresh version checked out from
CVS and i have not expected it will be so complicated to build it ...
something like 20+ steps (feels like hundreds of steps ...) to configure
AXIS-C++? this is insaneley complicated ..
why on earth there is a separate autogen/runconfig for ws-axis/c,
samples/client, and samples/server ?!
why configure script is not checked into CVS?! isnt it whole purpose of
configure script to ease that process?
why not have configure to correclty set apache1/2, expat/xercesc/both in
./configure intead of manually hacking
why there are so many warnings when compiling sampleS?
it seems that SOAPTransportFactory::initialize() has hardcoded
/usr/local/Axis/ so unless i change it i get this:
[aslom@archie bin]$ ./httpd -X
DLOPEN FAILED in loading transport library:
/usr/local/Axis/libs/libaxis_transport.so: cannot open shared object
file: No such file or directory
DLOPEN FAILED in loading transport library:
/usr/local/Axis/libs/libaxis_xmlparser.so: cannot open shared object
file: No such file or directory
Segmentation fault (core dumped)
currently it has this:
int XMLParserFactory::initialize()
{
#ifdef WIN32
#ifdef _DEBUG
m_pcLibraryPath = "AxisXMLParser_D.dll"; //this will be taken
from conf\
iguration file
//m_pcLibraryPath = g_pConfig->getAxConfProperty(AXCONF_XMLPARSER);
#else
m_pcLibraryPath = "AxisXMLParser.dll"; //this will be taken from
config\
uration file
//m_pcLibraryPath = g_pConfig->getAxConfProperty(AXCONF_XMLPARSER);
#endif
#else
m_pcLibraryPath = "/usr/local/Axis/libs/libaxis_xmlparser.so";
//this w\
ill be taken from configuration file
//m_pcLibraryPath = g_pConfig->getAxConfProperty(AXCONF_XMLPARSER);
#endif
if (!loadLib())
why //m_pcLibraryPath = g_pConfig->getAxConfProperty(AXCONF_XMLPARSER);
was commented?
so is everybody supposed to install /usr/local/Axis or nothing will work?
so i went and fixed it ti my /u/home/...tmp/Axis but here is the stopper
- it seems FileConfig is initialized with default constructor by
AxisConfig abut default constructor does nto initialize pFILEFile so
when is destructed out of struck it has non zero (random) in pFILEFile:
~AxisFile()
{
if(pFILEFile)
fclose(pFILEFile);
}
int AxisConfig::readConfFile ()
{
char carrLine[CONFBUFFSIZE];
char* pcValue;
AxisFile fileConfig; /*AxisFile is AxisC++ resource for file
manipulation */
char* sConfPath = NULL;
char* sNewConfPath = (char*) malloc(CONFBUFFSIZE);
char* key;
sConfPath = getenv ("AXIS_HOME");
m_pcValueArray[AXCONF_AXISHOME] = sConfPath;
if (!sConfPath)
return AXIS_FAIL;
m_pcAxisHome = (char*) malloc (CONFBUFFSIZE);
strcpy (m_pcAxisHome, sConfPath);
strcpy (sNewConfPath, sConfPath);
strcat (sNewConfPath, "/axiscpp.conf");
if (AXIS_SUCCESS != fileConfig.fileOpen(sNewConfPath, "r"))
return AXIS_FAIL;
while (AXIS_SUCCESS == fileConfig.fileGet(carrLine, CONFBUFFSIZE))
{
int linesize = strlen (carrLine);
pcValue = strpbrk (carrLine, ":");
if (!pcValue)
break;
key = (char*) malloc (strlen (pcValue));
pcValue[0] = '\0';
sscanf (carrLine, "%s", key);
if (strcmp (key, "WSDDFILEPATH") == 0)
{
m_pcValueArray[AXCONF_WSDDFILEPATH] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_WSDDFILEPATH], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_WSDDFILEPATH]+linesize - strlen
(key) - 2)
= '\0';
}
if (strcmp (key, "AXISLOGPATH") == 0)
{
m_pcValueArray[AXCONF_LOGPATH] = (char*) malloc (CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_LOGPATH], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_LOGPATH]+linesize - strlen (key) - 2)
= '\0';
}
if (strcmp (key, "AXISCLIENTLOGPATH") == 0)
{
m_pcValueArray[AXCONF_CLIENTLOGPATH] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_CLIENTLOGPATH], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_CLIENTLOGPATH]+linesize - strlen
(key) - 2)
= '\0';
}
if (strcmp (key, "CLIENTWSDDFILEPATH") == 0)
{
m_pcValueArray[AXCONF_CLIENTWSDDFILEPATH] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_CLIENTWSDDFILEPATH], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_CLIENTWSDDFILEPATH]+linesize -
strlen (key) - 2) = '\0';
}
if (strcmp (key, "AXISTRANSPORT_HTTP") == 0)
{
m_pcValueArray[AXCONF_TRANSPORTHTTP] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_TRANSPORTHTTP], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_TRANSPORTHTTP]+linesize -
strlen (key) - 2) = '\0';
}
if (strcmp (key, "AXISTRANSPORT_SMTP") == 0)
{
m_pcValueArray[AXCONF_TRANSPORTSMTP] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_TRANSPORTSMTP], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_TRANSPORTSMTP]+linesize -
strlen (key) - 2) = '\0';
}
if (strcmp (key, "AXISXMLPARSER") == 0)
{
m_pcValueArray[AXCONF_XMLPARSER] = (char*) malloc
(CONFBUFFSIZE);
strncpy (m_pcValueArray[AXCONF_XMLPARSER], pcValue +
1, linesize - strlen (key) - 2);
*(m_pcValueArray[AXCONF_XMLPARSER]+linesize -
strlen (key) - 2) = '\0';
}
}
return AXIS_SUCCESS;
}
so i get this lovely SEGFAULT:
(gdb) where
#0 0x4206b86a in fclose@GLIBC_2.0 () from /lib/tls/libc.so.6
#1 0x42061948 in fclose@@GLIBC_2.1 () from /lib/tls/libc.so.6
#2 0x4019a102 in AxisConfig::readConfFile() (this=0x80c87f8) at
AxisFile.h:103
#3 0x4017cf7e in initialize_module (bServer=1) at ../Axis.cpp:255
#4 0x40024bbc in module_init (svr_rec=0x80b3334, p=0x80b32fc)
at mod_axis.cpp:65
#5 0x0806dfce in ap_init_modules ()
#6 0x08079be4 in main ()
#7 0x42015704 in __libc_start_main () from /lib/tls/libc.so.6
(gdb) up
#1 0x42061948 in fclose@@GLIBC_2.1 () from /lib/tls/libc.so.6
(gdb) up
#2 0x4019a102 in AxisConfig::readConfFile() (this=0x80c87f8) at
AxisFile.h:103
103 fclose(pFILEFile);
Current language: auto; currently c++
(gdb) print *this;
Invalid character ';' in expression.
(gdb) print *this
$1 = {m_pcAxisHome = 0x0, m_pcKeyArray = {0x401bfea2 "WSDDFILEPATH",
0x401bfe7e "AXISLOGPATH", 0x401bfe8a "AXISCLIENTLOGPATH",
0x401bfe9c "CLIENTWSDDFILEPATH", 0x0, 0x401bfeaf "AXISTRANSPORT_HTTP",
0x401bfec2 "AXISTRANSPORT_SMTP"}, m_pcValueArray = {
0x401bfed5 "AXISXMLPARSER", 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}}
(gdb) print
$2 = {m_pcAxisHome = 0x0, m_pcKeyArray = {0x401bfea2 "WSDDFILEPATH",
0x401bfe7e "AXISLOGPATH", 0x401bfe8a "AXISCLIENTLOGPATH",
0x401bfe9c "CLIENTWSDDFILEPATH", 0x0, 0x401bfeaf "AXISTRANSPORT_HTTP",
0x401bfec2 "AXISTRANSPORT_SMTP"}, m_pcValueArray = {
0x401bfed5 "AXISXMLPARSER", 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}}
(gdb) print pFILEFile
so changing to iniitlaize calss variable (that is definitely not
cleared: it is in Java but no C++ and in C++ object allocated on stack
just take space AFAIK ...)
/**
* Default constructor
*/
AxisFile()
{
pFILEFile = 0;
}
do i miss something?
thanks,
alek
--
The best way to predict the future is to invent it - Alan Kay
--
The best way to predict the future is to invent it - Alan Kay