Speeding compilation on Solaris 8, Forte compiler 4.2 for C/C++

Hi,
The source code I'm compiling is a sequential compilation. The source code and binaries reside on internal disk.
The machine is a 4 CPU SUNV 440, 8GB RAM.
But still it takes 2..3 hours for compilation.
Any advise on improving the time factor will help.
Some parallel make option is not recommended here in the company :(
Thanks!
-Vikrant

since we have only internal disk, we have not copied the remote env & resources to the local disk
Will that matter?Depends. When it comes to performance problems its better to solve them one at a time. If you suspect your compilation system (compiler/make/build config) then you better take out network contention out of the equation.
-xjobsAny pointers?go to docs.sun.com and search for -xjobs, or just download the latest C++ User Guide.
Anyway -xjobs option first appears in C++ in Sun Studio 8, so C++ 4.2 does not have it yet.
You'll have to get a newer one (I do recommend to try FREE Sun Studio 11).
Would the current makefiles work as they are or there is a porting checklist?It depends on how you organize your build. IF all the dependencies are explicitly written in your makefiles then dmake will do a good job of handling it appropriately.
However if there are some implicit dependencies that are satisfied by a sequential order of your build then you can get a problem. It is easy to resolve though as soon as you realize where is the problem. You can always force a sequential order by using .NO_PARALLEL in your rule.
No other problems as far as I know.
As a bottom point - it seems strange that your compilation does not scale. You need to experiment with your build system to figure out what is the bottleneck.
regards,
Fedor.

Similar Messages

How to compile a dll for JNI in the CMD correctly

Now I find alot of old threads on using the mno command in cygwin to created dlls to use as an interface between c and Java, however I know this command is no longer available in cygwin and so I took the advid of downloading minGW and using this in the CMD instead.
However every example I try to compile and run this way throws the unsatified link error when the native function is to be called by the Java program. I believe it must be a compiling error creating a mismatch between the native function and the function to be called.
Here is the method I'm using to comile dlls for JNI, it is from the FAQ section of the minGw website:
gcc -Wall -D_JNI_IMPLEMENTATION_ -Wl,--kill-at
-Ic:/j2sdk1.4.1_02/include -Ic:/j2sdk1.4.1_02/include/win32
-shared someJavaImp.c -o JavaImp.dll
Is this how dll should be compiled? If not any suggestions?

sudsey wrote:
I know how to program well in C and Java, I just need to know the proper way to compile a dll for JNI.1. Get the MS IDE
2. Create a dll project - it must NOT be a managed dll
3. All done.

Compiling c++ files for windows

I have a c++ program that I would like to compile windows binaries for. I have tried installing mingw32-gcc, and using g++, but when I run the binaries (in wine or from windows), I get errors that certain .dll files are not found. Is there additional configuration needed to set up mingw32? Or should I be using a completely separate compiler. Thanks in advance!

ewaller@odin:~/test[1] 1040 %locate libgcc_s_sjlj-1.dll
/home/ewaller/Dropbox/libgcc_s_sjlj-1.dll
/home/ewaller/Dropbox/wxwidgetsprojects/spacewar/src/libgcc_s_sjlj-1.dll
/home/ewaller/devel/wxwidgetsprojects/spacewar/src/libgcc_s_sjlj-1.dll
/usr/i486-mingw32/bin/libgcc_s_sjlj-1.dll
/usr/i486-mingw32/lib/libgcc_s_sjlj-1.dll
ewaller@odin:~/test 1041 %pacman -Qo /usr/i486-mingw32/bin/libgcc_s_sjlj-1.dll
/usr/i486-mingw32/bin/libgcc_s_sjlj-1.dll is owned by mingw32-gcc 4.6.2-1
ewaller@odin:~/test 1042 %pacman -Qo /usr/i486-mingw32/lib/libgcc_s_sjlj-1.dll
/usr/i486-mingw32/lib/libgcc_s_sjlj-1.dll is owned by mingw32-gcc 4.6.2-1
ewaller@odin:~/test 1043 %
ewaller@odin:~/test 1054 %locate libstdc++-6.dll
/home/ewaller/Dropbox/libstdc++-6.dll
/home/ewaller/Dropbox/wxwidgetsprojects/spacewar/src/libstdc++-6.dll
/home/ewaller/devel/wxwidgetsprojects/spacewar/src/libstdc++-6.dll
/usr/i486-mingw32/bin/libstdc++-6.dll
/usr/i486-mingw32/lib/libstdc++-6.dll
ewaller@odin:~/test 1055 %pacman -Qo /usr/i486-mingw32/bin/libstdc++-6.dll
/usr/i486-mingw32/bin/libstdc++-6.dll is owned by mingw32-gcc 4.6.2-1
ewaller@odin:~/test 1056 %pacman -Qo /usr/i486-mingw32/lib/libstdc++-6.dll
/usr/i486-mingw32/lib/libstdc++-6.dll is owned by mingw32-gcc 4.6.2-1
ewaller@odin:~/test 1057 %
waller@odin:~/test 1047 %locate pdcurses.dll
ewaller@odin:~/test[1] 1048 %
Edit: Okay, I botched that post.. I'm fixing it.... There, that is better
Edit: The plot thickens...
ewaller@odin:~/test 1070 %pkgfile pdcurses.dll
ewaller@odin:~/test 1071 %
Last edited by ewaller (2012-03-23 04:35:19)

What are the compiler generated codes for objective-c?

After the compilation, the code for c/c++ is known as "object code" i.e. ".obj" file, and the binary executable file is ".exe" .
Similarly for java the compiled code is known as ".class" file.
In case of Objective-c what is the output of Compiler and what is the Executable file?

Strictly speaking, .obj and .exe refer only to the output of the compiler and linker for a Windows system (or cross compiler). For *nix compliant systems the output of the compiler is usually given a .o suffix and the executable produced after linking is named whatever you choose; but usually with no suffix at all.
Also, technically speaking, Java .class fiels don't fit neatly into this paradigm. The output of a Java compiler isn't object code, per se; it's optimized byte-code that can be run by the JRE.

Delphi doesn't compile the examples for DIO32HS

Delphi doesn't compile the examples for DIO32HS
Some libraries appear to be required,
but there are none supplied
According to the manual, one ActiveX should be sufficient.
What version is the source provided ?

The NI CD is Version 6.9.2
There are examples : \National Instruments\NI-DAQ\Examples\Delphi
There is no mention which Version of Delphi,
I have 3,5 and 6. However there should be components
to be installed into Delphi as DigitalWaveformContinous
requires TCWNumEdit, TCWDI and so on.
It also requires the units CWUIControlsLib_TLB,
CWAnalysisControlsLib_TLB, CWDAQControlsLib_TLB which are not part of Delphi istself but have to be provided
by NI.
Today I'm going to try Borland C++Builder, of which
I have version 4 and 5. This will also require some
components to be installed in BCB, I couldn't find them
yet.
\National Instruments\NI-DAQ\Examples\BorlandC
BTW: this ActiveX that was talked about couldn't be
found on the drive either. no *.ocx
It cannot be that hard. I just want a procedure
to setup the PCI-DIO-32HS, in terms of what is input
and output, how the handshaking is done and then
read/write to the port.
Rene

Can I compile an App for iOS from Dreamweaver CS5.5 from a PC running Windows (7)?

Can I compile an App for iOS from Dreamweaver CS5.5 from a PC running Windows (7)?

No, you need XCode to compile the app and you need a Developers license to be able to publish.

Compile jsp file for long time

Hi All,
I'm applying patch on EBS 11.5.9, now it's compiling jsp file for about 1hours.
Below is the log file,
enumerating jsps...13577
updating dependency...13
parsing jsp...13
writing deplist...13577
initializing compilation:
eliminating children...10257 (-3320)
searching uncompiled...13 (10 were stale)
translating and compiling:
searching untranslated...13
translating jsps...
Please advise.
Thanks,
Kevin

I'm applying patch on EBS 11.5.9, now it's compiling jsp file for about 1hours.
Below is the log file,What is the command you use to compile JSP file?
It is normal for compiling JSP files to run for more than 1 hour, which depends on the hardware resources you have. So, I would suggest you wait and see if the job completes successfully.
Thanks,
Hussein

Compile Add-On for SBO 2005 and 2007

Hi All,
I am looking for a document that I downloaded a while ago on steps how to compile add-ons for SBO version 2005 and 2007 on the same machine. I don't want to run both versions (I can be running either) just want to be able to compile add-ons on both version 2005 and version 2007 on my local machine. I also don't want to use virtual machine or something like that.
Thanks for the help,
Adele

Hi Adele,
usually it is enough to change the .dll reference in your code and recompile it.
you can also decide which diapi connection you use
when creating the company object
//The object OComp is defined as company object of the selected version
Dim OComp As SAPbobsCOM.Company
Set OComp = New SAPbobsCOM.Company
or
//The object OComp is defined as company object of version 2005
Dim OComp As SAPbobsCOM.Company
Set OComp = CreateObject("SAPbobsCOM.Company.6.8")
// for 2007 it should be
Dim OComp As SAPbobsCOM.Company
Set OComp = CreateObject("SAPbobsCOM.Company.8")

Can anybody compile CIFS modules for IdeaPhone P780

Can anybody be so kind to compile cifs modules for Lenovo IdeaPhone P780? It'll help us to connect network shares to use as local resources.
Kernel source for P780 I found here: http://mobilesupport.lenovo.com/us/en/products/phones/p-series/p780-smartphone
Thanks in advance

kleptophobiac wrote:Thanks a ton! Now what about madwifi?
uudecode ./../hal/linux/i386-elf.hal.o.uu
make[1]: uudecode: Command not found
make[1]: *** [hal.o] Error 127
You haven't got uudecode installed. It's part of the sharutils package, so install that.
- olly

Compile-time checking for Serializable

Hi,
For an object to be Serialized it must implement the Serializable interface. If any of the objects it stores as instance variables do not implement this interface then run-time exceptions will be thrown. Why can't there be a compile-time check for this? It seems fairly simple to me... Also, why don't the writeObject methods only accept objects that are Serializable? Just changing the method signature to public void writeObject(Serializable obj) throws... should be enough if compile-time checking were available.
Am I being really stupid or are Sun?
Dan.

It would be impossible to reasonably check at compile time, because what if you had a variable of type Object? At runtime it might work because anything could be in there, but if you check at compile time, it would have to fail, because the compiler can't gurantee that the value will be serializable. Also, a non-serializable field wouldn't cause a problem at runtime if it's null, which is another condition the compiler cannot check.

I have compiled a program for a target after I added a new vi from NI_AALPro.lvlib. The addition worked fine on my development computer but gave me errors when run on the target.

I have compiled a program for a target after I added a new vi from NI_AALPro.lvlib. The addition worked fine on my development computer but gave me errors when run on the target.
The box displayed:
Missing subVI NI_AALBase.lvlib:1D Linear Evaluation.vi in VI NI_PtbyPt.lvlib:Linear Fit PtbyPt.vi
Missing subVI NI_AALPro.lvlib:MSE.vi in VI NIPtbtPt.lvlib:Linear Fit PtbyPt.vi
What do I need to do to bring the correct library into the transportable code?

Hello,
I was unable to reproduce the behavior that you saw. I was able to deploy my application with the Linear Fit PtByPt without issues.
Could you attach the smallest piece of code that reproduces this behavior so that I can test it as well?
Justin Parker
National Instruments
Product Support Engineer

Solaris 10 SPARC Recommended Patch Cluster for 2008 quarter 3 version

Dear All,
Could any one please guide me, where I can download "Solaris 10 SPARC Recommended Patch Cluster" for 2008 quarter 3 version.
I have checked in sunsolve.sun.com, I'm able to find only the latest release.
please guide me.
Thanks and regards,
veera

Ok,
Here's a cute little formula to try using your systems parameters to gain some "head's-up" on an expected or estimated time to complete the patch run.
You will need a few things to prepare.
Number of "real" CPU's (not hyperthreads)
Speed of each CPU as a whole number i.e 2.87Ghz = 2,867
the total number of patches from the cluster i.e. Sept 15th = 381 (Solaris 10)
Network factor if using NFS = 2.5
local cluster file factor = 1.5
Patch cluster on CDROM factor = 3.25
Now, combine all of those elements in this equation:
{ ( #1 * #2 ) / #3 * factor(#4 or #5 or #6) }
This will yield a number in minutes of patch run time. Then all you will need to add is the standard boot up time to get the total of all patching and reboot.
Of course you could write a script to extract all of this information and feed it to "bc -l" to get a quick figure.
Example, on one of my Solaris 10 boxes with the information filled in to the equation:
#> echo "((4*2660)/381)*1.5"|bc -l
#> 41.88976377952755905511
It actually took 44.2 minutes to complete the patching of this box plus another 12 minutes to reboot. But all in all a pretty fair estimate I think.

Re: Running the same (Forte) application multiple times -for different

Hi
We had the same problem - how to deploy a number of identical applications, using each their own db.
(for training).
The solution we used is to wrap the entire application into different applications by using a very small
module called KURSUS01, KURSUS02 etc, that did nothing but call the start procedure of the main app.
Then in the dbsession connect, we made a call appname to get the application name, and appended the
first 8 chars to the dbname. Thus our dbnames now points to logicals name: rdbdataKURSUS01, rdbdataKURSUS02 etc.
All this allows us to deploy the identical apps in the same env, or change one version, and run both the old
and new program on the same pc and server at the same time (eg. KURSUS01 and KURSUS02).
I also think this is a kludge - but it works nicely!
Jens Chr
KAD/Denmark
-----Original Message-----
From: Haben, Dirk <[email protected]>
To: 'Soapbox Forte Users' <[email protected]>
Date: 15. januar 1999 09:41
Subject: Running the same (Forte) application multiple times - for different business clients.
Hi All
We have a number of different business clients all willing to use our
application.
The (forte) application is to run on our machines etc for these (business)
clients.
All (business) clients will have their data kept in separate Oracle DBs
(instance).
The problem now is that the entire (forte) application is written using
DBSessions.
Now, depending on what business client needs to be serviced (so to speak) we
need to attach to the right DB - or use the "right" SO.
The two options we can think of are:
Option1:
Programatic change to somehow "know" what (business) client (DB) I'm talking
about and then use the right DB.
Pro:
Only one forte environment to maintain
Can run multiple (business) clients on same PC at the same time
Con:
Requires many program changes
bending O-O rules(?)
can't dynamically name SOs so can it be done at all? (ResourceMGRs maybe?)
Option2:
Use separate environments! One for each business client.
Pro:
More defined separation of app and data,
SLA-easy
Con:
Maintain "n" number of environments
Can only run the application for one environment (business client) at a time
on one PC - Big Negative here!
Not knowing any feasible solution to option 1 (without much code changes and
developer moaning) I would go for option two; as I have already worked on
multi-environment setups on VMS back at the Hydro (hi guys).
I would appreciate any comments from anyone who has solved this problem.
How, Why Pro Con etc.
TIA,
Dirk Haben
Perth, WA
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Hi
We had the same problem - how to deploy a number of identical applications, using each their own db.
(for training).
The solution we used is to wrap the entire application into different applications by using a very small
module called KURSUS01, KURSUS02 etc, that did nothing but call the start procedure of the main app.
Then in the dbsession connect, we made a call appname to get the application name, and appended the
first 8 chars to the dbname. Thus our dbnames now points to logicals name: rdbdataKURSUS01, rdbdataKURSUS02 etc.
All this allows us to deploy the identical apps in the same env, or change one version, and run both the old
and new program on the same pc and server at the same time (eg. KURSUS01 and KURSUS02).
I also think this is a kludge - but it works nicely!
Jens Chr
KAD/Denmark
-----Original Message-----
From: Haben, Dirk <[email protected]>
To: 'Soapbox Forte Users' <[email protected]>
Date: 15. januar 1999 09:41
Subject: Running the same (Forte) application multiple times - for different business clients.
Hi All
We have a number of different business clients all willing to use our
application.
The (forte) application is to run on our machines etc for these (business)
clients.
All (business) clients will have their data kept in separate Oracle DBs
(instance).
The problem now is that the entire (forte) application is written using
DBSessions.
Now, depending on what business client needs to be serviced (so to speak) we
need to attach to the right DB - or use the "right" SO.
The two options we can think of are:
Option1:
Programatic change to somehow "know" what (business) client (DB) I'm talking
about and then use the right DB.
Pro:
Only one forte environment to maintain
Can run multiple (business) clients on same PC at the same time
Con:
Requires many program changes
bending O-O rules(?)
can't dynamically name SOs so can it be done at all? (ResourceMGRs maybe?)
Option2:
Use separate environments! One for each business client.
Pro:
More defined separation of app and data,
SLA-easy
Con:
Maintain "n" number of environments
Can only run the application for one environment (business client) at a time
on one PC - Big Negative here!
Not knowing any feasible solution to option 1 (without much code changes and
developer moaning) I would go for option two; as I have already worked on
multi-environment setups on VMS back at the Hydro (hi guys).
I would appreciate any comments from anyone who has solved this problem.
How, Why Pro Con etc.
TIA,
Dirk Haben
Perth, WA
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Solaris 8 x86 doesn't ask for default route at install time

The Solaris 8 installation program doesn't ask what the machine's default route should be. This means that when you configure DNS servers later on that aren't on the same subnet as the machine you get error messages.
Anyone know a work around?

Re: Solaris 8 x86 doesn't ask for default route at install time
I had the same problem.
I was wondering why something so essential is missing for the GUI Install, it would be possilbe a patch to correct this?, instead of configuring nsswitch, and another files.
The easier the install and admin, more users solaris will have on Intel Platform.
P.D. Sorry for my english.
05-sep-00 : 0:51 :GavinCam
The Solaris 8 installation program doesn't ask what the machine's default route should be. This means that when you configure DNS servers later on that aren't on the same subnet as the machine you get error messages.
Anyone know a work around?

What happended to the Solaris Crash Analysis Tool 4.2 for Solaris 10?

What happended to the Solaris Crash Analysis Tool 4.2 for Solaris 10?
I'm trying to debug a kernel panic under Solaris 10 on sparcv9. I tried the SUNWscat 4.1 package but it core dumps on Solaris 10. I found a blog entry that described the problem and mentioned a version 4.2 fix. That was from May 2006, but I have not found any update sense then. Was 4.2 ever released or is their an alternative tool for Solaris 10 that I can use?

One problem is that those of us who develop Solaris CAT do so as a "hobby" as it's not part of our actual jobs. But he good news is that management gave us the time to get a release out and 5.0 is now available here . We're now planning 6 month releases so we don't have a 5 year delay between updates.
BTW, the Solaris Crash Analysis Tool supports Solaris 10, OpenSolaris/Solaris Nevada, sparc, and x86/x64.
There are so many enhancements and bug fixes that I'd recommend reading the release notes (SUNWscat/docs/index.html).
-- Dana

Re: (forte-users) Session management for page builder(fwd)

Jaco,
Hope this helps,
John
John Soper, Information Systems Development, ITS, The University of Melbourne
email: j.soperits.unimelb.edu.au >>>> Tel: 9344 5612---------- Forwarded message ----------
Date: Mon, 10 Jan 2000 16:34:31 +1100
From: Lyle Winton <L.Wintonits.unimelb.edu.au>
To: John Soper <j.soperits.unimelb.edu.au>
Subject: Re: (forte-users) Session management for page builder (fwd)
Why not construct an intermediate page after the
login page that has SESSION_UNSPECIFIED and
a refresh META tag. The page can then refresh
to either the login failed or login succeeded pages
depending on how the login went! Looks like...
1) Login page (SESSION_UNSPECIFIED)
2A) Refresh page (SESSION_UNSPECIFIED)
< HTML >
< HEAD >
< META http-equiv="refresh"
content="0;URL=<a href=
"http://www.blah.com/forte.cgi?PageName=3">http://www.blah.com/forte.cgi?PageName=3</a>" >
< /HEAD >
< BODY >
Login succeeded. Please wait...
< /BODY >
< /HTML >
2B) Refresh page (SESSION_UNSPECIFIED)
< HTML >
< BODY >
Login failed.
< /BODY >
< /HTML >
3) We're finally in. (SESSION_REQUIRED)
I'm not sure if this works on internet exploder.
Lyle.
John Soper wrote:
Lyle,
(Post from forte mailing group)
Does this make sense to you?
John
John Soper, Information Systems Development, ITS, The University of Melbourne
email: j.soperits.unimelb.edu.au >>>> Tel: 9344 5612---------- Forwarded message ----------
Date: Thu, 30 Dec 1999 07:54:24 +0200
From: "Jaco Erasmus (home)" <jacoerasmweb.co.za>
To: kamranaminyahoo.com
Subject: (forte-users) Session management for page builder
Hi everybody,
We have a lot of legacy code making use of the page builder service to
produce web pages. These pages were originally written without session
management. I'm now busy adding session management to them, but there is
one problem with this approach and I will appreciate if someone can shed
some light on it. Here it is:
Page one is submitted.
Some validation (authentication) takes place and depending on the outcome,
either page 2A (SESSION_REQUIRED) or 2B (error page with
SESSION_UNSPECIFIED) must be displayed. In order to implement this, I
needed a place to make a decision. The way I've done it, is to pass a
'virtual page' (SESSION_UNSPECIFIED) to the page builder service. The
validation is done here and request.PageName is then replaced with the
PageName of pages 2A or 2B. The HandleRequest() method is then called
again. The problem is that the ValidateSession() method does not get
invoked again, thus allowing 2A through without a session. How do I make
sure that the ValidateSession() method get invoked again?
The approach making use of templates look to me as if it has all the means
to do this (redirect tag), but I don't want to rewrite everything if I
don't have to. Is there a way that a pagebuilder page can be specified by
the redirect tag? This will definitely help, but so far I've only managed
to call templates from the redirect tag.
Is the template approach better suited for session management? It is
definetely better documented...
Regards.
Jaco
For the archives, go to: http://lists.sageit.com/forte-users and use
the login: forte and the password: archive. To unsubscribe, send in a new
email the word: 'Unsubscribe' to: forte-users-requestlists.sageit.com

Hi,
i hope this helps
http://help.sap.com/saphelp_nw70/helpdata/EN/7e/aa610cc1dd8f4388b1df02fc362f0f/frameset.htm
http://help.sap.com/saphelp_nw70/helpdata/EN/69/c250754ba111d189750000e8322d00/frameset.htm
regards,
Anil.

Speeding compilation on Solaris 8, Forte compiler 4.2 for C/C++

Similar Messages

Maybe you are looking for