[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#442338: Fwd: Bug#442338: FTBFS: error: 'CURLE_FTP_USER_PASSWORD_INCORRECT' was not declared in this scope



Hi,

Daniel Stenberg wrote:
>> curl 7.17.0 breaks the build of openoffice.org. Rene has already found the 
>> cause, it is described below.
>
> Eh, no. curl 7.17.0 does *not* break this as you will see below.

Incorrect. It breaks. As you say yourself.

>> Is it only a formal error by the curl side or openoffice.org requires to 
>> be upgraded to the new API?
>
> The error code CURLE_FTP_USER_PASSWORD_INCORRECT has been moved to the list 
> of codes that will be removed in the future, and thus it isn't present if 
> you define CURL_NO_OLDIES when you build.
>
> So if you want backwards-compatible style you *don't* define CURL_NO_OLDIES 
> and you can use that define fine still. If you define it, you should be 
> prepared to edit some of the codes to the up-to-date versions.
>
>>>> | g++ -fmessage-length=0 -c -O2 -fno-strict-aliasing -DCURL_NO_OLDIES 
>>>> -I.
>
> So why do these guys explicitly switch off compatibility?

This was done between 1.12 and 1.13 (so pre 2.0.0 beta even) of ucb/source/ucp/makefile.mk
with the following in cvs log:

revision 1.13
date: 2004/09/20 14:32:42;  author: rt;  state: Exp;  lines: +3 -1
branches:  1.13.16;  1.13.66;
INTEGRATION: CWS systemcurl1 (1.12.54); FILE MERGED
2004/09/20 14:17:24 cmc 1.12.54.1: #i34405# curl has changed its headers
slightly

The Issue URL is http://www.openoffice.org/issues/show_bug.cgi?id=34405 and it seems to be a
part of a commit to fix builds with system-curl and which says:
"
   Description: Opened: Mon Sep 20 14:15:00 +0000 2004 Sort by: Oldest first | Newest first
     __________________________________________________________________________________________________________________________________________________

--with-system-curl and configure doesn't work with latest curl seeing as curl
moved from CURLOPT_PASSWDFUNCTION being 66 to being 0 which was covered, and now
its moved to -3 by default. Also the curl/types.h is deprecated in favour of
curl/curl.h. Changes allow both old and new CURL varients to work.
"

Which indeed only does this:

diff -u -u -r1.2 -r1.2.186.1
--- source/ucp/ftp/ftphandleprovider.hxx        15 Oct 2002 09:21:17 -0000      1.2
+++ source/ucp/ftp/ftphandleprovider.hxx        20 Sep 2004 14:17:23 -0000      1.2.186.1
@@ -1,5 +1,5 @@
 #include <rtl/ustring.hxx>
-#include <curl/types.h>
+#include <curl/curl.h>

 namespace ftp {

diff -u -u -r1.5 -r1.5.188.1
--- source/ucp/ftp/ftploaderthread.hxx  28 Aug 2002 07:23:14 -0000      1.5
+++ source/ucp/ftp/ftploaderthread.hxx  20 Sep 2004 14:17:23 -0000      1.5.188.1
@@ -71,9 +71,7 @@
 #ifndef _OSL_THREAD_H_
 #include <osl/thread.h>
 #endif
-#ifndef __CURL_TYPES_H
-#include <curl/types.h>
-#endif
+#include <curl/curl.h>


 namespace ftp {
diff -u -u -r1.12 -r1.12.54.1
--- source/ucp/ftp/makefile.mk  2 Feb 2004 18:36:22 -0000       1.12
+++ source/ucp/ftp/makefile.mk  20 Sep 2004 14:17:24 -0000      1.12.54.1
@@ -75,7 +75,9 @@

 .INCLUDE: settings.mk

-#CFLAGS +=-fno-inline
+.IF "$(SYSTEM_CURL)" == "YES"
+CFLAGS+=-DCURL_NO_OLDIES
+.ENDIF

 # --- General -----------------------------------------------------

This is only added for builds with "system-curl". Stock OOo as provided
by Sun uses an internal curl copy. 7.12.2. No, no typo.. Amd anything which is done for
the ftp ucp needs to build/work with both.

Maybe it was simply a security measure? RH did that, so I don't know what he intended with it.
(Cced).

>>> It seems that curl wanted to keep compatibility but actually broke it 
>>> because the subsitution doesn't work?
>
> Nope. We still support the option as described above.

Not when you define CURL_NO_OLDIES. 

Grüße/Regards,

René
-- 
 .''`.  René Engelhard -- Debian GNU/Linux Developer
 : :' : http://www.debian.org | http://people.debian.org/~rene/
 `. `'  rene@debian.org | GnuPG-Key ID: 248AEB73
   `-   Fingerprint: 41FA F208 28D4 7CA5 19BB  7AD9 F859 90B0 248A EB73

Attachment: signature.asc
Description: Digital signature


Reply to: