Mailing-List: contact cygwin-apps-help AT cygwin DOT com; run by ezmlm Sender: cygwin-apps-owner AT cygwin DOT com List-Subscribe: List-Archive: List-Post: List-Help: , Delivered-To: mailing list cygwin-apps AT cygwin DOT com From: "Gary R. Van Sickle" To: Subject: RE: has anyone tried latest setup.exe from cvs ? Date: Sat, 15 Dec 2001 00:02:32 -0600 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) In-Reply-To: <3C1A1012.CD77BEEF@syntrex.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 Importance: Normal > -----Original Message----- > From: cygwin-apps-owner AT cygwin DOT com > [mailto:cygwin-apps-owner AT cygwin DOT com]On Behalf Of Pavel Tsekov > Sent: Friday, December 14, 2001 8:44 AM > To: cygwin-apps AT cygwin DOT com > Subject: has anyone tried latest setup.exe from cvs ? > > > Has anyone ran this beast successfully ? > Sort of. The problem though is that I'm nearing the completion of some pretty extensive changes to the GUI code, so if there's problems with that in the cvs stuff I might not even know about it. I did however run across a problem in the INI-parsing code that does prevent it from working. The problem is with entries in setup.ini that look like this: @ jbigkit sdesc: "Lossless image compression library" ldesc: "JBIG is a highly effective lossless compression algorithm for bi-level images (one bit per pixel), which is particularly suitable for document pages." category: Graphics Libs requires: cygwin i.e., that have no "version:" lines in them (what is such an entry supposed to mean, or is this actually a upset bug?). The parser then never creates a version entry in the corresponding packagemeta, and hence you get a GPF in packagemeta::SDesc() when it tries to "return versions[1]->SDesc ();". I've been getting around this with the following changes, which AFAICT are in accordance with Robert's comment, "/* SDesc is global in theory, across all package versions. [...]". All I know for sure is that these keep choose.cc from SEGFAULTing, and don't appear to interfere with anything (I don't get a jbigkit listing though): Index: package_meta.h =================================================================== RCS file: /cvs/src/src/winsup/cinstall/package_meta.h,v retrieving revision 2.7 diff -p -u -b -r2.7 package_meta.h --- package_meta.h 2001/12/02 03:25:11 2.7 +++ package_meta.h 2001/12/15 05:49:34 @@ -85,6 +85,7 @@ public: char *installed_from; /* SDesc is global in theory, across all package versions. LDesc is not: it can be different per version */ + char *sdesc; char const *SDesc (); /* what categories does this package belong in. Note that if multiple versions * of a package disagree.... the first one read in will take precedence. Index: package_meta.cc =================================================================== RCS file: /cvs/src/src/winsup/cinstall/package_meta.cc,v retrieving revision 2.7 diff -p -u -b -r2.7 package_meta.cc --- package_meta.cc 2001/12/02 03:25:11 2.7 +++ package_meta.cc 2001/12/15 06:02:08 @@ -151,5 +151,6 @@ packagemeta::add_category (Category & ca char const * packagemeta::SDesc () { - return versions[1]->SDesc (); + //return versions[1]->SDesc (); + return sdesc; }; Index: iniparse.y =================================================================== RCS file: /cvs/src/src/winsup/cinstall/iniparse.y,v retrieving revision 2.20 diff -p -u -b -r2.20 iniparse.y --- iniparse.y 2001/12/02 03:25:11 2.20 +++ iniparse.y 2001/12/15 06:06:33 @@ -94,7 +94,7 @@ lines simple_line : VERSION STRING { cpv->set_canonical_version ($2); add_correct_version ();} - | SDESC STRING { cpv->set_sdesc ($2); } + | SDESC STRING { cp->sdesc = $2; cpv->set_sdesc ($2); } | LDESC STRING { cpv->set_ldesc ($2); } | CATEGORY categories | REQUIRES requires -- Gary R. Van Sickle Brewer. Patriot.