From: Whit B. <wh...@tr...> - 2004-07-26 18:00:29
|
Hi all, Figured it was time to start with a virgin nightly again, but dang: --- Loading up virgin wiki %BODY% Fatal error: Call to a member function on a non-object in /web/closeread/phpwiki/lib/Template.php(133) : eval()'d code on line 5 --- - which looks like the same spot several of us make bug reports on to the list well over a month ago. Since I followed the INSTALL and mysql.INSTALL files fairly closely (which until a couple of months ago was a foolproof way to go), um, what's not going right? Ah, switching ALLOW_ANON_USER from true to false gets beyond this - to a "You Must Sign In to Read this Page" screen. Fiddling with other stuff in the auth configuration doesn't seem to resolve this, as long as ALLOW_ANON_USER is true. Does someone have a set of configuration settings which work with the current CVS which include ALLOW_ANON_USER=true and ALLOW_ANON_EDIT=false and ALLOW_BOGO_LOGIN=false, and which enforces a true login method (not bogo) for those with edit access? I know there are still some deep issues with the CVS version, but basic stuff like this should be working now, right? Is something critical missing from the INSTALL instructions, or is the user auth code still this broken? Thanks, Whit |
From: Dan F. <dfr...@cs...> - 2004-07-26 18:11:17
|
Whit Blauvelt wrote: >Hi all, > > > Figured it was time to start with a virgin nightly again, but dang: Seems good. >Since I followed the INSTALL and mysql.INSTALL files fairly closely (which >until a couple of months ago was a foolproof way to go), um, what's not >going right? > > Don't know. >Ah, switching ALLOW_ANON_USER from true to false gets beyond this - to a >"You Must Sign In to Read this Page" screen. Fiddling with other stuff in >the auth configuration doesn't seem to resolve this, as long as >ALLOW_ANON_USER is true. > >Does someone have a set of configuration settings which work with the >current CVS which include ALLOW_ANON_USER=true and ALLOW_ANON_EDIT=false and >ALLOW_BOGO_LOGIN=false, and which enforces a true login method (not bogo) >for those with edit access? > > Haven't checked. >I know there are still some deep issues with the CVS version, but basic >stuff like this should be working now, right? Is something critical missing >from the INSTALL instructions, or is the user auth code still this broken? > > I think Reini is really the right person to answer this, and he's gone until early August. Dan |
From: Whit B. <wh...@tr...> - 2004-07-26 19:07:32
|
On Mon, Jul 26, 2004 at 01:11:13PM -0500, Dan Frankowski wrote: > >Does someone have a set of configuration settings which work with the > >current CVS which include ALLOW_ANON_USER=true and ALLOW_ANON_EDIT=false > >and > >ALLOW_BOGO_LOGIN=false, and which enforces a true login method (not bogo) > >for those with edit access? > Haven't checked. > > >I know there are still some deep issues with the CVS version, but basic > >stuff like this should be working now, right? Is something critical missing > >from the INSTALL instructions, or is the user auth code still this broken? > > I think Reini is really the right person to answer this, and he's gone > until early August. Dan - Thanks for the response. Everyone - Since I asked about this same stuff about six weeks ago, and Reini responded to something else in my query but ignored the primary queries, and since there are others experimenting with the code ... please if you've gotten it to run with this seemingly-legal set of configuration options - especially if you learned something about working around the problem I'm running into - post to the list. Maybe it can work as a draft of an improved section for the INSTALL files? Thanks! Whit |
From: Dan F. <dfr...@cs...> - 2004-07-26 19:15:42
|
Whit Blauvelt wrote: >On Mon, Jul 26, 2004 at 01:11:13PM -0500, Dan Frankowski wrote: > > > >>>Does someone have a set of configuration settings which work with the >>>current CVS which include ALLOW_ANON_USER=true and ALLOW_ANON_EDIT=false >>>and >>>ALLOW_BOGO_LOGIN=false, and which enforces a true login method (not bogo) >>>for those with edit access? >>> >>> > > > >>Haven't checked. >> >> >> >>>I know there are still some deep issues with the CVS version, but basic >>>stuff like this should be working now, right? Is something critical missing >>> >>> >>>from the INSTALL instructions, or is the user auth code still this broken? >> >>I think Reini is really the right person to answer this, and he's gone >>until early August. >> >> > >Dan - > >Thanks for the response. > >Everyone - > >Since I asked about this same stuff about six weeks ago, and Reini >responded to something else in my query but ignored the primary queries, and >since there are others experimenting with the code ... please if you've >gotten it to run with this seemingly-legal set of configuration options - >especially if you learned something about working around the problem I'm >running into - post to the list. Maybe it can work as a draft of an improved >section for the INSTALL files? > > Good idea. I did not say (but should have) that it's great that you're testing and agitating for improvement. PhpWiki needs a legion of folks like you .. and of course a legion of developers too! :-) >Thanks! > >Whit > > Likewise thanks. Dan |
From: <ru...@x-...> - 2004-07-28 07:43:32
|
> Whit Blauvelt wrote: >>I know there are still some deep issues with the CVS version, but basic >> stuff like this should be working now, right? Is something critical >> missing from the INSTALL instructions, or is the user auth code still >> this broken? User auth is only somehow broken for HttpAuth, everything else should work ok. The deep issues in CVS to be solved are only in the mysql database iterators andConvertOldMarkup(). > I think Reini is really the right person to answer this, and he's gone > until early August. Well, I'm not gone. I'm in Poland until 2.Aug, and away from my machine, so I cannot fix anything or answer difficult questions. -- reini |
From: Whit B. <wh...@tr...> - 2004-07-26 21:43:19
|
On Mon, Jul 26, 2004 at 02:00:25PM -0400, Whit Blauvelt wrote: > Ah, switching ALLOW_ANON_USER from true to false gets beyond this - to a > "You Must Sign In to Read this Page" screen. Fiddling with other stuff in > the auth configuration doesn't seem to resolve this, as long as > ALLOW_ANON_USER is true. Just a note: Checked the obvious thing: With it true, after that login screen, still the same failure: --- Loading up virgin wiki %BODY% Fatal error: Call to a member function on a non-object in /web/closeread/phpwiki/lib/Template.php(133) : eval()'d code on line 5 --- Then trying switching to File rather than SQL for data type, that login page has --- PHP Warnings lib/DbSession.php:49: Warning[512]: Your WikiDB DB backend 'file' cannot be used for DbSession. Set USE_DB_SESSION to false. --- Well, USE_DB_SESSION isn't even in the current config.ini (and it's not a php.ini thing either). And adding "USE_DB_SESSION = false" to config.ini _still_ presents this error message, and logging in still presents a failed next page, which also includes that DbSession warning. Is the warning obsolete? The real fix something else? If anyone has started with a current virgin CVS copy and been skilled or lucky enough to set it up so it worked, maybe if you shared your config.ini (or whatever you guess might be the most pertinent settings) we could begin to narrow down the real, working options for current use. That Reini is most concerned yet with things that make it go crash is good priorities; but meanwhile I'm after a working beta of the latest code to test some modifications on, so that when Reini goes stable I can just add those in and be ready to roll. Thanks again, Whit |