Next stable release (roadmap + question) (kados, read carefully)
Hello all, I think we have almost solved the 2 main problems with rel_2_2 : * encoding problems : joshua get rid of unicode ones & I get rid of the remaining iso-8859-1 one. I made something i'm not very proud of : Joshua introduced a TemplateEncoding system preference that can be either iso8859-1 or utf8. If it is set to iso8859-1, there are a lot of encoding problems : MARC::File::XML don't like it, as MARC::Record is supposed to work in MARC8 or UNICODE. Thus, it seems that all accents are transformed in the process html => xml => marc. Thus, i've reintroduced the html => marc behaviour if the TemplateEncoding is set to iso8859-1 (i've fixed the html2marc as joshua proposed previously, fixing a remaining bug) This solution is a workaround : we will have to find a correct solution for Koha 3.0, to move every Koha to UNICODE and UNICODE only. But for instance, I tried some things but failed. Thust this solution, that works correctly for french libraries and should change nothing for unicode ones.' 'acqui.simple/addbiblio.pl' 2>&1 * MARC editor, cloneTag problem : when duplicating a tag, there was 2 problems : - the mandatory flags where no more properly affected. I solved this problem by moving color change on Check to the field itself, not to error$i container. Error$i container is now useless and removed. I also changed active color to grey, yellow being for mandatory fields. - when you duplicate a field with an authority, the value is still reported to the 1st field. When you duplicate the 1st authority, the container name is set to index+index, so you get 2 "indexindex" containers. I still have to solve this problem, but that should not bee too hard I think. Kados : I didn't modified NPL templates, I let you do this. ROADMAP : ========== I hope I'll find some more time to work on this version next week, working on translation to french, and maybe install it to one of my libraries. So, this next version should be officially released on 12th / 13th. what could be it's name ? 2.2.6 or 2.4.0 ? That will be discussed on monday (note for everybody : monday is closed in France. So you won't see us on chat. But i'll be here on monday 20:00GMT (10PM for French ppl) -- Paul POULAIN et Henri Damien LAURENT Consultants indépendants en logiciels libres et bibliothéconomie (http://www.koha-fr.org)
On Fri, Jun 02, 2006 at 05:45:11PM +0200, Paul POULAIN wrote:
I think we have almost solved the 2 main problems with rel_2_2 :
* encoding problems : joshua get rid of unicode ones & I get rid of the remaining iso-8859-1 one. I made something i'm not very proud of :
Joshua introduced a TemplateEncoding system preference that can be either iso8859-1 or utf8. If it is set to iso8859-1, there are a lot of encoding problems : MARC::File::XML don't like it, as MARC::Record is supposed to work in MARC8 or UNICODE. Thus, it seems that all accents are transformed in the process html => xml => marc. Thus, i've reintroduced the html => marc behaviour if the TemplateEncoding is set to iso8859-1 (i've fixed the html2marc as joshua proposed previously, fixing a remaining bug) This solution is a workaround : we will have to find a correct solution for Koha 3.0, to move every Koha to UNICODE and UNICODE only. But for instance, I tried some things but failed. Thust this solution, that works correctly for french libraries and should change nothing for unicode ones.' 'acqui.simple/addbiblio.pl' 2>&1
This got me thinking about a new Unicode tutorial I read last week: http://juerd.nl/perlunitut.html It describes perl support for encoding and decoding characters. It gave me a clearer picture of what encodings are, and how to work with them in perl. A very worthwhile read. -kolibrie
Hi all, On Fri, Jun 02, 2006 at 05:45:11PM +0200, Paul POULAIN wrote:
I think we have almost solved the 2 main problems with rel_2_2 :
* encoding problems : joshua get rid of unicode ones & I get rid of the remaining iso-8859-1 one. I made something i'm not very proud of :
Joshua introduced a TemplateEncoding system preference that can be either iso8859-1 or utf8. If it is set to iso8859-1, there are a lot of encoding problems : MARC::File::XML don't like it, as MARC::Record is supposed to work in MARC8 or UNICODE. However, Mike Rylander introduced a UNIMARC-handling feature in the latest version of MARC::File::XML (on SF) that will not attempt to convert to/from MARC8. I urge you to attempt this solution, as I belive the XML parsers shouldn't have any trouble dealing with 8859 encoded data.
Also, I'm still waiting to hear back from Tumer on whether the M::F::X stuff, specifically libxml2 and XML::LibXML as a parser, is working on the Windows platform when dealing with combining characters (it works on Linux now with the proper versions of things; I detailed my findings in an earlier email and created some tests in HEAD C4/tests/Record-test.pl that can be used to check your system; these tests should be expanded to include UNIMARC records and encodings, something I will try to do this weekend if I have a chance).
Thus, it seems that all accents are transformed in the process html => xml => marc. Thus, i've reintroduced the html => marc behaviour if the TemplateEncoding is set to iso8859-1 (i've fixed the html2marc as joshua proposed previously, fixing a remaining bug) There were several bugs, which did you fix?:
* saves blank subfields * subfield order is hardcoded to always start with 'a' for repeatable tags (because it is hardcoded in the addfield routine). * only possible to specify one set of indicators for each set of tags (ie, one for all the 650s). (because they are stored in a hash with the tag as the key). * the underlying routines don't support subfield reordering or subfield repeatability.
This solution is a workaround : we will have to find a correct solution for Koha 3.0, to move every Koha to UNICODE and UNICODE only. But for instance, I tried some things but failed. Thust this solution, that works correctly for french libraries and should change nothing for unicode ones.' 'acqui.simple/addbiblio.pl' 2>&1
* MARC editor, cloneTag problem : when duplicating a tag, there was 2 problems : - the mandatory flags where no more properly affected. I solved this problem by moving color change on Check to the field itself, not to error$i container. Error$i container is now useless and removed. I also changed active color to grey, yellow being for mandatory fields. - when you duplicate a field with an authority, the value is still reported to the 1st field. When you duplicate the 1st authority, the container name is set to index+index, so you get 2 "indexindex" containers. I still have to solve this problem, but that should not bee too hard I think.
Kados : I didn't modified NPL templates, I let you do this. OK, let me know when you've found a solution and either Owen or I will update NPL ... thanks!
ROADMAP : ========== I hope I'll find some more time to work on this version next week, working on translation to french, and maybe install it to one of my libraries.
So, this next version should be officially released on 12th / 13th.
what could be it's name ? 2.2.6 or 2.4.0 ? That will be discussed on monday (note for everybody : monday is closed in France. So you won't see us on chat. But i'll be here on monday 20:00GMT (10PM for French ppl) I'm planning on turning dev-week into a 2.4 release, just as soon as it's stable enough (chris and I are planning to do a 2.3 release sometime this weekend or early next week). Since rel_2_2 doesn't really have any _new features_, just bug fixes (I know we disagree on this point), I think it should be released as 2.2.6. However, I wouldn't mind seeing a 2.2.5 version (unstable) release beforehand just to give some folks time to find bugs and us time to fix them before the official 'stable' release. Anyway, for what it's worth, that's my opinion. We can certainly discuss it more on Monday.
Cheers, -- Joshua Ferraro VENDOR SERVICES FOR OPEN-SOURCE SOFTWARE President, Technology migration, training, maintenance, support LibLime Featuring Koha Open-Source ILS jmf@liblime.com |Full Demos at http://liblime.com/koha |1(888)KohaILS
Joshua Ferraro a écrit : > Hi all, > > On Fri, Jun 02, 2006 at 05:45:11PM +0200, Paul POULAIN wrote: >> I think we have almost solved the 2 main problems with rel_2_2 : >> >> * encoding problems : joshua get rid of unicode ones & I get rid of the >> remaining iso-8859-1 one. I made something i'm not very proud of : >> >> Joshua introduced a TemplateEncoding system preference that can be >> either iso8859-1 or utf8. >> If it is set to iso8859-1, there are a lot of encoding problems : >> MARC::File::XML don't like it, as MARC::Record is supposed to work in >> MARC8 or UNICODE. > However, Mike Rylander introduced a UNIMARC-handling feature in the latest > version of MARC::File::XML (on SF) that will not attempt to convert to/from > MARC8. I urge you to attempt this solution, as I belive the XML parsers > shouldn't have any trouble dealing with 8859 encoded data. I installed MARC-XML v0.83 from CPAN, and have the problem. Should I try the CVS version ? >> Thus, it seems that all accents are transformed in the process html => >> xml => marc. >> Thus, i've reintroduced the html => marc behaviour if the >> TemplateEncoding is set to iso8859-1 >> (i've fixed the html2marc as joshua proposed previously, fixing a >> remaining bug) > There were several bugs, which did you fix?: > * saves blank subfields > * subfield order is hardcoded to always start with 'a' for repeatable tags (because it is hardcoded in the addfield routine). > * only possible to specify one set of indicators for each set of tags (ie, one for all the 650s). (because they are stored in a hash with the tag as the key). > * the underlying routines don't support subfield reordering or subfield repeatability. I Just call html2marc (the one you fixed before switching to html2xml) instead of html2xml then xml2marc It seems to work fine for me. >> * MARC editor, cloneTag problem : >> when duplicating a tag, there was 2 problems : >> - the mandatory flags where no more properly affected. I solved this >> problem by moving color change on Check to the field itself, not to >> error$i container. Error$i container is now useless and removed. I also >> changed active color to grey, yellow being for mandatory fields. >> - when you duplicate a field with an authority, the value is still >> reported to the 1st field. When you duplicate the 1st authority, the >> container name is set to index+index, so you get 2 "indexindex" containers. >> I still have to solve this problem, but that should not bee too hard I >> think. >> Kados : I didn't modified NPL templates, I let you do this. > OK, let me know when you've found a solution and either Owen or > I will update NPL ... thanks! OK, i'll let you know for sure. > Since rel_2_2 doesn't really have any > _new features_, just bug fixes (I know we disagree on this point), yes ;-) > I think > it should be released as 2.2.6. However, I wouldn't mind seeing a 2.2.5 > version (unstable) release beforehand just to give some folks time to > find bugs and us time to fix them before the official 'stable' release. > Anyway, for what it's worth, that's my opinion. We can certainly discuss > it more on Monday. Sounds a good plan to me : what I want to avoid is having a stable release that is unstable. So I would probably release a 2.2.6RC1 or something like that, that i'll install to one of my libraries. -- Paul POULAIN et Henri Damien LAURENT Consultants indépendants en logiciels libres et bibliothéconomie (http://www.koha-fr.org)
participants (3)
-
Joshua Ferraro -
Nathan Gray -
Paul POULAIN