$url
For the record, I've committed dev-libs/libxls and dev-libs/quazip in tree, new opt. deps of qtiplot.
bump could be found here: http://dev.gentooexperimental.org/~jlec/qtiplot/
Thank you :) In portage
Since I've worked on this package too, I'm going to change ebuild after I test all the changes I did. Still I have some questions I hope, somebody who worked on this ebuild could answer. Justin could you tell me, please? 1. Why pygsl is in dep, where it's used? 2. edos2unix why is that necessary? 3. # the lib$$suff did not work in the fitRational*.pro files Is this true? I failed to reproduce any failure yet. Thats are things I failed to find answers. Still there are many other issues that are still in ebuild and I hope are fixed here: svn co http://overlays.gentoo.org/svn/dev/pva/sci-visualization/qtiplot and after I finish tests in most USE flag combinations I'm going to add newer revision. Comments to work in progress are welcome. )
(In reply to comment #4) > 1. Why pygsl is in dep, where it's used? I only added new deps to the ebuild, but didn't remove old ones. So it can be that an older versions needed pygsl. But as the package uses gsl and has a python interface it somehow makes sense to hav pygsl, but I don't know where thisdep comes from. > 2. edos2unix why is that necessary? I should have commented that on the ebuild. Som of the files to patch have CRLF line ends, so modifying patches is evil, because the line ends will probably not fit anymore. I added this to make writting and maintaining patches simpler. > 3. # the lib$$suff did not work in the fitRational*.pro files Ancient, didn't test if that could be removed. > Is this true? I failed to reproduce any failure yet. > > Thats are things I failed to find answers. > > Still there are many other issues that are still in ebuild and I hope are fixed > here: > > svn co http://overlays.gentoo.org/svn/dev/pva/sci-visualization/qtiplot > > and after I finish tests in most USE flag combinations I'm going to add newer > revision. Comments to work in progress are welcome. ) >
We should merge the ebuilds: You dropped the sip-4.8 patch which is needed EMF support could be fixed with the emfengine ebuild in the sci overlay and knowledge about const char* conversion errors. I personally like clean patches more than this seding like you did, but that's just a matter of taste. So the ebuilds principly are doing the same, just in an different way, except the fitRational.pro patching. Are you sure the optional python stuff works? With .10 there was some problems with that.
* ERROR: sci-visualization/qtiplot-0.9.7.11 failed: * multilib-strict check failed! * * Call stack: * misc-functions.sh, line 804: Called install_qa_check * misc-functions.sh, line 526: Called die * The specific snippet of code: * [[ ${abort} == yes ]] && die "multilib-strict check failed!"
Created attachment 215783 [details] /var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/temp/build.log build.log with USE="python" and sip-4.8 with your ebuild
(In reply to comment #5) > But as the package uses gsl and has a python interface it somehow makes sense > to hav pygsl, but I don't know where thisdep comes from. Ok, I'll keep this then. > > 2. edos2unix why is that necessary? > Som of the files to patch have CRLF line ends, so modifying patches is evil, Using vim, there is nothing evil in different CRLFs ;) At least I haven't seen any problems. (In reply to comment #6) > We should merge the ebuilds: True, but since you've just updated patches, and I've recreated most of them from scratch to reflect current ebuild I guess I just incorporate everything that became missed. > You dropped the sip-4.8 patch which is needed Yea, very likely. That's what I was going to check first. And in any case I'd better put >=sip-4.9 and dropped that patch. > EMF support could be fixed with the emfengine ebuild in the sci overlay and > knowledge about const char* conversion errors. That's what I had in mind. If you manage to write some patch I'll be happy to include that. > Are you sure the optional python stuff works? With .10 there was some problems > with that. The problem was it did not build, and now it does here. Thank you for help.
(In reply to comment #8) > build.log with USE="python" and sip-4.8 with your ebuild Oh, true. This saved me ab ti of time. I'll update deps then.
Peter, some more things to fix: We need x11-libs/gl2ps[png,zlib] and define LIBPNG_LIBS or make it optional. Also defintion of SYS_LIB and SYS_INCLUDE should be in the ebuild for easy import to gentoo prefix
(In reply to comment #9) > (In reply to comment #5) > > But as the package uses gsl and has a python interface it somehow makes sense > > to hav pygsl, but I don't know where thisdep comes from. > Ok, I'll keep this then. If I remember correctly, pygsl was recommended for python scripting. Suppose I want to plot some formula with special functions; this can be done with a python script, and pygsl has many special functions in it.
qtiplot-0.9.7.11.ebuild which is currently in the tree fails for me with USE=doc: .... make[1]: Leaving directory `/var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/qtiplot' make docbook2html -d qtiplot.dsl docbook-en/index.docbook -e no-valid Using catalogs: /etc/sgml/xml-docbook-4.4.cat Using stylesheet: /var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/manual/qtiplot.dsl Working on: /var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/manual/docbook-en/index.docbook jade:/usr/share/sgml/docbook/xml-dtd-4.4/ent/isogrk4.ent:42:30:E: "1D6C2" is not a character number in the document character set <many similar messages skipped> jade:E: cannot open "/usr/share/sgml/stylesheet/dsssl/modular/html/docbook.dsl" (No such file or directory) jade:E: specification document does not have the DSSSL architecture as a base architecture jade:/var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/manual/./help.dsl:408:7:E: "entity" is not the name of any flow object class jade:/var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/manual/./help.dsl:92:7:E: reference to undefined variable "htmlnewline" <many similar messages skipped> jade:/var/tmp/portage/sci-visualization/qtiplot-0.9.7.11/work/qtiplot-0.9.7.11/manual/./help.dsl:273:37:E: reference to undefined variable "sgml-root-element" january 2007Version updated for QtiPlot 0.9.7.8The QtiPlot HandbookIonVasiliefion_vasilief@yahoo.frRogerGadiouRoger.Gadiou@uha.frKnutFrankeKnut.Franke@gmx.de2004 - 2009Ion Vasilief2006 - june 2007Roger Gadiou and Knut FrankeLegal notice: Permission is granted to copy, distribute and/or modify <a long text from docs follows in the log> the pixel intensity versus pixel index.How can I make a graph that has one x-axis, but two y-axes with different orders of magnitude? Select at least two columns that you want to display and use the Double-Y command. make: *** [web] Error 8 * ERROR: sci-visualization/qtiplot-0.9.7.11 failed: * html docbook building failed * * Call stack: * ebuild.sh, line 48: Called src_compile * environment, line 3932: Called die * The specific snippet of code: * emake || die "html docbook building failed";
Andrey, and have you tried ebuild from my overlay?
(In reply to comment #11) > Peter, > some more things to fix: > We need x11-libs/gl2ps[png,zlib] and > define LIBPNG_LIBS or make it optional. Where is that used? As I see it only defines GL2PS_HAVE_LIBPNG which is used in gl2ps (that we build separately). !isEmpty(LIBPNG_LIBS) { DEFINES += GL2PS_HAVE_LIBPNG INCLUDEPATH += $$LIBPNG_INCLUDEPATH LIBS += $$LIBPNG_LIBS } > Also defintion of SYS_LIB and SYS_INCLUDE should be in the ebuild for easy > import to gentoo prefix Well, much more things are required for prefix, so I'll let prefix guys to handle this. In any case I think the way I'm doing now is much easier to customize. I'm going to open bug upstream to avoid QTeXEngine and liborigin patches, and rename translations to something more consistent with other projects. That said, updated version is in the tree. Please report bugs, improvements if necessary.
Thanks peter
As already mentioned in comment #7 your ebuild fails with FEATURES="multilib-strict" which should be respected.
Created attachment 215916 [details] /var/tmp/portage/sci-visualization/qtiplot-0.9.7.11-r1/temp/build.log FEATURES="multilib-strict" failed build.log
(In reply to comment #17) > As already mentioned in comment #7 your ebuild fails with > FEATURES="multilib-strict" which should be respected. Ah, I guess that is why lib$$suff was needed. Well, I need to check if better solution exist. Thanks.
Created attachment 215918 [details] missing icons And I am missing some icons. I also tried starting qtiplot with "-d" to use a black profile with same result.
I am missing all toolbar icons, too. Changing the language from Spanish to English didn't solve the problem.
Ok, multilib and icons issue are fixed in 0.9.7.11-r2. Thank you guys.
(In reply to comment #22) > Ok, multilib and icons issue are fixed in 0.9.7.11-r2. Thank you guys. > Thanks _you_!!
