Topic: Building headless Bitcoin and Bitcoin-qt on Windows - page 41. (Read 419389 times)

Fixing the makefile seems to solve the linking error

LIBS= \
 -l boost_system-mgw48-mt-s-1_5x \
 -l boost_filesystem-mgw48-mt-s-1_5x \
 -l boost_program_options-mgw48-mt-s-1_5x \
 -l boost_thread-mgw48-mt-s-1_5x \

I am trying to compile smartcoind.exe, but  I am bumping in the following error

C:/deps/boost_1_53_0/boost/program_options/eof_iterator.hpp:75: undefined refere
nce to `boost::program_options::detail::common_config_file_iterator::get()'
obj/util.o: In function `ZN5boost15program_options6detail26basic_config_file_ite
ratorIcEC1ERSiRKSt3setISsSt4lessISsESaISsEEb':
C:/deps/boost_1_53_0/boost/program_options/detail/config_file.hpp:145: undefined
 reference to `boost::program_options::detail::common_config_file_iterator::comm
on_config_file_iterator(std::set, std::alloc
ator > const&, bool)'
C:/deps/boost_1_53_0/boost/program_options/detail/config_file.hpp:148: undefined
 reference to `boost::program_options::detail::common_config_file_iterator::get(
)'
collect2.exe: error: ld returned 1 exit status
makefile.mingw:76: recipe for target 'smartcoind.exe' failed
mingw32-make: *** [smartcoind.exe] Error 1


Anybody experience this before?

If this was a linux box, I'd say you didn't have the QtGUI package installed.  Have you tried this with QT 5.2.0 per post #1?

Cassey

Note that your output says it is looking for gcc 4.6.2 and you say you are using gcc 2.6.2??? So something is wrong there for starters!

Also, traditionally we use BerkeleyDB 4.8.30 note db 4.6.3!

Ron

I had exactly the same problem, as did several others.   Solved it by going back to post #1 in this thread, which has been routinely updated, and following the instructions EXACTLY - use all the same packages and releases as specified in post #1.

I am currently trying to compile litecoin-qt.exe using makefile.debug but keep getting this error:


I think its something to do with Qt but I cant figure out what... Any ideas?

I am using
boost 1.55
db 4.6.3
mingw 2.6.2
openssl-1.0.1c
qt 4.8.5


I am able to compile using makefile.release but it just runs for a few seconds with no GUI and then stops again.

Nvm, fixed it.  Apparently when they say add the config file AFTER it tells you that you don't have one, they really do mean only AFTER.

Set permissions/ issue the fix that keeps it from complaining about them.  Nitro wrote this somewhere in the guide, but you can get this fix by googling it I think.  I'll post the exact solution for this later. 

everytime I set  a genesis block for the daemon and change the nNonce, the epoch time, etc, rebuilding and re-running the coind.exe says the merkle root assertion failed.  The thing works perfectly in test net...

i compiled the bitoind.exe and is synchronizing( i can see through the log file in the roaming folder is getting the blocks)

i dont really understand how from one guide are coming different problem and different solution but here is mine:

about the winsock.h winsock2...

in ws2tcpip.h i found this part and seems it's already using winsock2.h and works ( ) so i removed that check...


scrolling down about the struct ip_mrq already declared i just commented the declaration:


to solve the last error i removed this:





EDIT:

Trying to compile Qt(step 4.2) i get this error:

Go into your mingw install and replace "winsock.h" with a copy of "winsock2.h".

I tried a variety of ways to force it to be included and none worked for reasons I don't understand, but this brute force approach did!

tried this yesterday night, but got stuck in this exact same point/error... if anybody got any suggestion

What makes you think it has anything to do with Windows?
I use Qt 4.8.4, gcc 4.6.2 (unobtanium now!), Boost 1.53, BerkeleyDB 4.8.30, OpenSSL 1.0.1c or 1.0.1e and levelDB 1.13 and build bitcoin with no problems.  I can't speak to litecoin as I don't know when its sources "forked" off of bitcoin.  It is important to know what version of bitcoin, and so what date, that an alternate crypto currency was born, as the tool versions that were used at the time would be most appropriate for that alt coin.  As the alt coin "matures" one may update the tools, but they need to be updated in a "gradual" "coordinated" fashion.  For example, one wouldn't - shouldn't use too much "newer" gccs as their "behavior" is known to "destabilize" "older" Qts for example.

Have you tired the tool combinations that fail somehow on Windows, and build the same alt coin (or bitcoin) on Linux (or MacOS) and test to see if they work there?

The main problem people have in building bitcoind & bitcoin-qt on Windows is that they aren't meticulous (enough) in following the recipe of nitrogenetics.  

Of course, it helps if you are knowledgeable on Windows!  Especially on command boxes (aka DOS windows).  It helps greatly if you know how to use .bat files or .wsh files, that is, how to "do" batch programming.  Also it helps if you are knowledgeable on how the Windows path environment is kept, controlled and manipulated.  There are many copies of the path at any moment in Windows, you know!  

There are subtle issues of pathnames with spaces in them, though that is less of an issue now than it was in the past.  And also one must be aware of subtleties regarding "\" versus "/" in couching pathnames.  None of these are difficult issues, it is just that the knowledge has been lost since there are no longer any definitive books on these subjects, and the general Windows user is no longer expected to know or wants to know, how  to do these things.  Remember that Windows 7 and 8 now make it even more difficult to "get your fingers in there and meddle around".

It helps if one knows the history of Windows and its DOS heritage.  Without that knowledge you have to fumble around by analogy with whatever you know about Linux, Unix, MacOs, etc. and  try to discern the "truth" of any subtle question by internet "research" and have the good judgement to know what is true, half-true or pure BS.  Kind of a vicious circle (catch-22) in which the only way to get knowledge is to know enough already to be able to tell what is knowledge

So if you think the bitcoind (and Boost) code is pure and clean, try building it without the gcc -w argument!  Let's not even talk about qmake, makefile.release/debug and bitcoin-qt, which doesn't have the -w argument and you know what it "spews forth"

What's interesting is that the compiler "warnings" about the bad C++ slang that is being used, as opposed to the preferred slang, i.e. the highly idiomatic, baroque (as Knuth calls it) nature of C++, can be controlled in MSVC++ and one can "see" more clearly what is "going on".  The thought is that this should lead to more understanding, but so far, the answers given are swamped by the questions that now appear!  Which is why you haven't heard from me (yet) on the Windows MSVC2005 version of bitcoind.  I have a "debug" version of bitcoind 0.8.6 that "appears to" run correctly.  And yes, Claire123 has a MSVC2012 version of bitcoind (& bitcoin-qt), but do our versions work correctly?  Do we have 'release' versions?  I am hoping to "GitHub" commit-pull request my MSVC "additions" to the bitcoind sources soon, but only when I feel that the code runs reliably.  

All of this reminds me of the Zen quote for today:
The most dangerous thing in the world is to think you understand something.

Ron

You're welcome
Ron

I'm not sure if building the rest of the environment, like QT (which came later in the instruction), or if installing the precompiled boost binaries fixed it, but the problem did go away.

Having problems building boost:

Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation.  All rights reserved.

C:\Users\cassey>cd c:\deps

c:\deps>cd boost*

c:\deps\boost_1_55_0>bootstrap.bat mingw
Building Boost.Build engine
'"VCVARS32.BAT"' is not recognized as an internal or external command,
operable program or batch file.
The system cannot find the batch label specified - Test_Option
'"VCVARS32.BAT"' is not recognized as an internal or external command,
operable program or batch file.
'"VCVARS32.BAT"' is not recognized as an internal or external command,
operable program or batch file.
'cl' is not recognized as an internal or external command,
operable program or batch file.

Update (and trimmed post):  I tried installing the pre-compile binaries by the final link failed.  Went back and tried building 1_53_0 and it worked fine.  For hells bell, tried it again in the 1_55_0 and the problem disappeared.  Strange, but maybe it will help someone else via Google searching in the future...

Update2:  After rebuilding, my coin linked clean - and for the VERY FIRST TIME on windows, actually worked!  Thank you SO MUCH for maintaining this thread!

 

The exception unknown software () in the application at location ()

Click on OK to terminate the program.

I managed to create altcoin.exe i altcoin-qt but when you run show me this error. Any advice?




translate window

The exception unknown software () in the application at location ()

Click on OK to terminate the program.

You may shoot ME a message or an e-mail Address is in my forum signature.

Oh god.... maybe I should try out my own currency for a bit on different systems before releasing it?  I know you are probably busy, but you mind helping out, Solcoin Project?  I would not be opposed to compensating you.  Also, you did get me to look into solcoin with your name, so maybe that's something I should have done with this account (or with another one?).  Anyway, I have a couple of questions to ask you not related to qt or BTC, but trading on Cryptsy.  Shoot me a private message, but preferably an email @ [email protected].  This offer is open to anyone who has Dev experience as well, Solcoin Project has just been particularly amiable and helpful.



PS: I don't think I'm a newbie, according to the standards set by the forum in terms of knowledge, but I definitely don't have a presence here.

We wrestled quite a while with Windows when building the Qt 4.8 wallet for our currency Solcoin, forked from Litecoin 0.8.6.1. Depending on the versions of GCC, the Boost libraries, OpenSSL and especially BDB, the results varied greatly. The client would run perfectly on one Windows installation but fail on another.

The two recurring problems were that the wallet software either refused to start up because of failure to initialize the database, even in the case of creating a brand new such, or it would crash upon exit. We finally got the build right using GCC 4.6.3 (POSIX threading model) instead of 4.4.0 or 4.8.2, Boost 1.55, OpenSSL 1.0.1c, BDB 4.8.30 and our own build of the Qt 4.8.5 libraries, and finally linking all of it statically.