Re: web pages
Hi Mikhael, On 15 Apr 2003 at 01:49:54 +, Mikhael Goikhman wrote: By meta information I mean the multiline header in all php files. For example the line: //- Last modified : 15.04.2003 08:36:53 uwp This belongs to the revision control system, it manages it automatically without a need to store this meta information in the file body itself. You are right. I haven't think on the header in the files. My emacs updates this header automatically each time I save a buffer. I have removed this line in all files. This fancy feature is already done, isn't it? All .B FvwmTaskBar or .I FvwmTaskBar are replaced with a link to its man page. I thought I have found some examples where this link was not made. For example in the FvwmButtons Manpage there is a FvwmPager which is not a link. $this_site = logo competition; ^^ The value of $this_site is used as name of a php array. So the contents of this_site must be a valid identifier according to programming languages. Yes, but I didn't use a space here, I used logo_competition_rules and logo_competition names, I meant to use the last one in both cases. No worries, was just a note. I have thought to remove this restriction anyway by using a two-dimensional array. Are you doing the man page conversion for the new web site? Regards, Uwe -- ,_, Uwe Pross (O,O) mailto:[EMAIL PROTECTED] ( ) http://www.tu-chemnitz.de/~uwp =--- -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
On 15 Apr 2003 08:37:37 +0200, Uwe Pross wrote: Are you doing the man page conversion for the new web site? Yes, today or most probably tomorrow. Regards, Mikhael. -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
Hi Mikhael, On 15 Apr 2003 at 09:10:13 +, Mikhael Goikhman wrote: Are you doing the man page conversion for the new web site? Yes, today or most probably tomorrow. That's good. To support passing of different layouts between fvwm-sites relative links should be done like: a href=?php echo conv_link_target('stable/fvwm_man.php');?fvwm/a This uses the php-function conv_link_target which must be defined in the actual layout file. Please check if your pages propagate the actual layout definition to other pages: http://www.fvwm.org/documentation/manpages/index.php?layout=plain I am going to write some developer information on the new website about the layout interface, html-conventions, image names, Hopefully I find the time to do that today. Before that I am going to finish the entry documentation page. This might be done tonight. When the man pages and the doc page are ready, we might put the first version online ;-) Regards, Uwe -- ,_, Uwe Pross (O,O) mailto:[EMAIL PROTECTED] ( ) http://www.tu-chemnitz.de/~uwp =--- -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
On 15 Apr 2003 11:31:52 +0200, Uwe Pross wrote: On 15 Apr 2003 at 09:10:13 +, Mikhael Goikhman wrote: Are you doing the man page conversion for the new web site? Yes, today or most probably tomorrow. That's good. To support passing of different layouts between fvwm-sites relative links should be done like: a href=?php echo conv_link_target('stable/fvwm_man.php');?fvwm/a This uses the php-function conv_link_target which must be defined in the actual layout file. Please check if your pages propagate the actual layout definition to other pages: http://www.fvwm.org/documentation/manpages/index.php?layout=plain I think it will be much more simple for us and more convenient/consistent for a user if we don't use/pass GET parameters to define all kinds of layouts, themes, css-s. Yes, I really mean this. Only use session values that are valid for the entire site per user. Users with write cookies turned off only get the default layout or the last stored layout (some users enable cookies temporarily only to store settings once). This is a good practice. This way, external sites may link to: http://fvwm.org/documentation/manpages/stable/ and different users will see it differently depending on their previous settings. I am really concerned about how external and internal links look. I don't like very much if the links include any GET parameters. Of course you may optionally implement a logic that if GET parameters are defined they are used (override) the user settings. But by default no url should contain these parameters, i.e. normally the user layout should not be passed in the link. When the man pages and the doc page are ready, we might put the first version online ;-) Probably on April 20 if everything goes well. Someone should write announcements. Regards, Mikhael. -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
Hi Mikhael, On 15 Apr 2003 at 10:24:45 +, Mikhael Goikhman wrote: I think it will be much more simple for us and more convenient/consistent for a user if we don't use/pass GET parameters to define all kinds of layouts, themes, css-s. Yes, I really mean this. Only use session values that are valid for the entire site per user. Users with write cookies turned off only get the default layout or the last stored layout (some users enable cookies temporarily only to store settings once). I don't see any bad issues by passing parameters via url. This is not mandatory - It is a feature. I have discussed with Bob about using cookies about two weeks ago and we agreed to implement it when we have fot the time. For now GET parameters are implemented and can be used. An automatic selection using cookies should not be that difficult to set up. I have planned to make a separate page which contains all available layouts from which the user may decide if he wants do set a cookie. I don't like to force the user to enable cookies. This is a good practice. This way, external sites may link to: http://fvwm.org/documentation/manpages/stable/ and different users will see it differently depending on their previous settings. A user which comes from a external site gets the default layout. If he changes the default settings he may bookmark it or does it again at next time he views the pages. I am really concerned about how external and internal links look. I don't like very much if the links include any GET parameters. Changing the layout and keep it changed when viewing subpages is a feature I haven't seen on any other web site. With the parameters in the url the user can see how this is been done. Of course you may optionally implement a logic that if GET parameters are defined they are used (override) the user settings. But by default no url should contain these parameters, i.e. normally the user layout should not be passed in the link. The layout defs are only passed in the link if the user has changed it. Anyway, I think even if we don't use parameter passing via url it is good to use the wrapper function conv_link_target since it makes it possible to manipulate relative links inside the php-scripts. This function can just return its argument or append layout definitions as we wish. Further one could generate a html version of the php-tree by changing the link target from *php to *html and rip the whole tree using wget. When the man pages and the doc page are ready, we might put the first version online ;-) Probably on April 20 if everything goes well. It should be possible to get all pages ready be this time. But I won't have any time between 18 and 21 April. Removing index.html in the root directory should be the only thing to do. Maybe we can put a .htaccess file in the root directory which says the server should serve index.php before index.html, so we can keep the old version fully. I don't know if the server supports this. Regards, Uwe -- ,_, Uwe Pross (O,O) mailto:[EMAIL PROTECTED] ( ) http://www.tu-chemnitz.de/~uwp =--- -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
On 15 Apr 2003 15:46:55 +0200, Uwe Pross wrote: On 15 Apr 2003 at 12:34:48 +, Mikhael Goikhman wrote: Suppose someone wants to post (or include in his site) a link to the FvwmButtons man page. If he copies it and it has ?layout=something, all users are forced to read it using this specific layout that some may find unreadable. This is bad. Now, suppose he is smart enough to post a link with everything after ? stripped. Now all users are forced to see the default layout. This is not the best thing to do either. The best possible way IMO is not to have any GET parameters in url and use stored session values only (if any). How about somebody wants to put a link to the button page with exact a specific layout? I think we should support both. By default the user will be asked to set a cookie to store a layout definition. But we may offer the possibility to access different layout via GET parameters. Maybe the user is asked to type this url by hand. If GET parameters are given we can assume that the user wants a specific layout and use it instead of default or cookie settings. In that case it may be useful to pass this parameters to sub pages. (We may also hide the parameters from the url and store it in a cookie automatically. But I don't like that that much.) I think that this is the best thing to do. One more thing. It would be nice to have 404.php (configured in apache) that shows an error and suggests to go to the home page. It may also include a match table that maps the old url to the new one to suggest it. A possible auto redirection may be used, but this is arguable. Does the server return the file /404.php if the requested page is not on the server? I have to read some apache docu to learn how this works. It returns the result of whatever is configured in ErrorDocument 404. It may be configured to use a redirection or a local substitution (this differs in which url of the missing page a user gets). I prefer a substitution, i.e. /generated/FAQ.html shows the result of /error/404.php Regards, Mikhael. -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
MG == Mikhael Goikhman [EMAIL PROTECTED] writes: MG It returns the result of whatever is configured in ErrorDocument MG 404. It may be configured to use a redirection or a local MG substitution (this differs in which url of the missing page a user MG gets). I prefer a substitution, i.e. /generated/FAQ.html shows the MG result of /error/404.php You can put ErrorDocument 404 /some/url in a .htaccess file and check that in; it should work. - J -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
On 10 Apr 2003 08:43:09 +0200, Uwe Pross wrote: On 09 Apr 2003 at 21:09:23 +, Mikhael Goikhman wrote: No, I didn't mean files that are fully generated, only files that are mostly written by hand but contain lines (meta information) that should be updated on every file change. Do you mean the actual version numbers of fvwm for example? By meta information I mean the multiline header in all php files. For example the line: //- Last modified : 09.04.2003 08:14:03 uwp This belongs to the revision control system, it manages it automatically without a need to store this meta information in the file body itself. I planed to extend the current so that certain keywords on the files will be converted to links to other man pages. For example all occurences of FvwmTaskBar will be replaced by a link to the FvwmTaskBar manpage. This is done in all manpages except the FvwmTaskBar manpage itself. But this a fancy feature and can wait ;-) This fancy feature is already done, isn't it? All .B FvwmTaskBar or .I FvwmTaskBar are replaced with a link to its man page. (Should we shorten documentation to doc?) You can do that if you like. Currently, it is 50/50 for me. Both are good... BTW: One thing I forgot to mention yesterday. I saw that you used a identifier containing spaces in your php file like: $this_site = logo competition; ^^ The value of $this_site is used as name of a php array. So the contents of this_site must be a valid identifier according to programming languages. Yes, but I didn't use a space here, I used logo_competition_rules and logo_competition names, I meant to use the last one in both cases. Regards, Mikhael. -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
Hi Mikhael, On 09 Apr 2003 at 21:09:23 +, Mikhael Goikhman wrote: No, I didn't mean files that are fully generated, only files that are mostly written by hand but contain lines (meta information) that should be updated on every file change. Do you mean the actual version numbers of fvwm for example? This may be obtained automatically from fvwm-cvs/ftp or we place a meta information file in the tree where some php-variables are set like: ?php $stable_version = 2.4.16 $unstable_version = 2.5.6 ? This file could be included in the php files as needed. Ideally, all meta information should be outside of the file contents, but of course there may be exceptions, for example some web pages may contain text This page was last updated on XXX as part of the content. In the actual version the last update time is placed on the bottom of the page. This time is obtained from the timestamp of the php file to be displayed. So no need to place update information in the php itself. For example, I don't yet understand why sometimes lines inside a window-based decorations are wrapped as if there is a table that is 30% or similar of the whole window decoration width. I realized this yesterday when I started porting the module information page. I fixed it this morning. The table inside the window decoration did not have a width attribute. I have set it to 100% which means the contents uses 100% of the windows width. Also there is a broken layout and uninformative error when there are consistency problems in $parent and $this on different pages. But you probably fixed this. Navgen fixed this problem silently before. Now a warning is given if the definitions in parent and child does not match. Also I removed an error message on the web site if the actual paged does not belong to the navigation array. In the actual version the value of $this_page is set to the first element of the $top_array which is the home page. This new behaviour avoids displaying unecessary error messages. I think you refer to the old php. Let's just suppose everyone has a new php and thus a command line php utility, so granting any write access to the web server is not needed. Okay. If you think this is not too much restriction I will try to make navgen run in the CLI environment. The problem with this is that you should first commit and only after some time (it is updated once in several hours) check out fvwm.org or fvwm.sf.net, because links like file://... or http://localhost/; can't be checked. Is there a validator that accepts input (say, as a form field) instead of url? This is not urgent and not very important at all. Yes, I got the same problems when I wrote sites at home. The tidy program is even more restrictive than the web validator and can be used with wget like: wget -q -O - http://localhost/fvwm-web/download/index.php | tidy -errors -q will show you errors on the file. You may want to hide the links to the validiation pages for this request since they produce warnings. The links are inserted from layout_default.inc and may be not inserted when the request comes from localhost or your host name. Please check this files the links are hidden for the web checkers already. You can check which php-variable can be used to hide the links by viewing the file php_info on your webserver and find out what is specific to your server. I planned only to convert html to php, i.e. to continue with man2html. If you think it is not good enough, you may try tidy. Just, as you said the current script does dozens of small fixes of both .1 and the resulting .html, so the output becomes correct and usable. For the first time we can use the current script and run tidy afterwards. After this we need to replace the header and footer of the html output and replace it with a php header and php footer. I wrote templates for this. You could add the tidy run and the header/footer-exchange to the current script. I planed to extend the current so that certain keywords on the files will be converted to links to other man pages. For example all occurences of FvwmTaskBar will be replaced by a link to the FvwmTaskBar manpage. This is done in all manpages except the FvwmTaskBar manpage itself. But this a fancy feature and can wait ;-) I also planned to have this structure: documentation/manpages - links to stable and unstable, and to perllib documentation/manpages/stable - contains man pages of 2.4.x documentation/manpages/unstable - contains man pages of 2.5.x documentation/perllib - links to stable and unstable, and to manpages documentation/perllib/stable - contains man pages of 2.4.x documentation/perllib/unstable - contains man pages of 2.5.x Good idea. No worries. I would like to remove the perlib documentation from the module information site and place it separately. (Should we shorten documentation to doc?) You can do that if you like. I actually prefer long names for variables and
Re: web pages
Hi Mikhael, On 09 Apr 2003 at 02:55:10 +, Mikhael Goikhman wrote: Autogenerated lines in cvs files is not a good idea. What do you mean by auto? In the new version there is not and it was not planed to autogenerate anything. In the new version some file for example author information or latest news *may* be created from the corresponding files (AUTHOR, NEWS) in the fvwm tree. I have provided awk-scripts for this. For example autogenerated Last-Modified-Date The new version of the layout file inserts the last modified date of the actual file automatically. The time is taken from the file's timestamp on the webserver. or Author or History only lead to unnecessary cvs conflicts (if 2 people work on different parts of the same file). Since those files are created from the corresponding files of the fvwm tree, the conflicts should be not too big. Similarly about cvs variables like $Id, $Log. Agree $Id, $Log should not apear in the web files. The last character of text files in cvs should better be EOLN, otherwise a probability of cvs conflicts (unnecessarily) grows. Please force whole lines. I don't see the point here, but I think emacs can do that automatically. I will have a look how this can be done. BTW, in some new pages I used a style that we use for our C/Perl code, i.e. one TAB for indentation, always a space after a comma and so on. For indentation I use corresponding modes in emacs. It indents text automatically. I will check the files according to this style conventions. Finally, I really like a progress in moving to a new design. Converting the old html-pages to a w3c-conform html takes time. Some old pages are far away for standard html: http://www.mirror.ac.uk/services/validator/check?uri=http%3A%2F%2Fwww-user.tu-chemnitz.de%2F%7Euwp%2Ffvwm-web%2Fcvs.html I don't like to put broken html on the new page. I just don't understand yet everything and it seems there are bugs. I hope not ;-) I did not have time to make a short documentation how the new php concept works (except a mail addressed to parv and fvwm-workers a couple of weeks ago) and how new sites are added/modified on the new web tree. I will do that in a while. Just some notes here: - You need a php web server which you grants write access to the file navgen_write/navigation.inc - After designing a new page you need to run navgen.php on your web server to update the navigation array which is stored in the file navgen_write/navigation.inc. navgen.php reports warnings and errors on the site structure. With no errors and warnings the output of navgen should look like http://www-user.tu-chemnitz.de/~uwp/fvwm-web/navgen.php The new version of php provides a command line interface which allows to process php-files without a web server. To use this CLI with navgen some adaptions are neccessary since the environment is different. Im going to change navgen so that it runs on a web server and on the CLI. - After cleaning the problems reported by navgen please check your new site with the html and css checker. It will tell you what's not html/css-conform on your site. BTW: Both of your news sites passed this check :-) If you just make a new page and commit it without running navgen. Your site won't work as expected in most cases. I plan to convert manpages and perllib documentation. Actually I am working on the creation of the manpage files at the moment. At least yesterday night ;-) The problem is that man2html produces non standard html. I found a tool http://tidy.sourceforge.net/ which can clean this html afterwards. I planed to write a new script or rewrite the actual script which makes some further adaption to get a nice manpage files for the fvwm web site. Since I might not have time till weekend you could do the manpage conversion script. Please let me know if you are going to do it so that I can send you some dones, ideas and plannings. Regards, Uwe -- ,_, Uwe Pross (O,O) mailto:[EMAIL PROTECTED] ( ) http://www.tu-chemnitz.de/~uwp =--- -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
On 09 Apr 2003 09:57:36 +0200, Uwe Pross wrote: On 09 Apr 2003 at 02:55:10 +, Mikhael Goikhman wrote: Autogenerated lines in cvs files is not a good idea. What do you mean by auto? In the new version there is not and it was not planed to autogenerate anything. In the new version some file for example author information or latest news *may* be created from the corresponding files (AUTHOR, NEWS) in the fvwm tree. I have provided awk-scripts for this. No, I didn't mean files that are fully generated, only files that are mostly written by hand but contain lines (meta information) that should be updated on every file change. Ideally, all meta information should be outside of the file contents, but of course there may be exceptions, for example some web pages may contain text This page was last updated on XXX as part of the content. But this text at least serves some purpose, not is just a comment. I just don't understand yet everything and it seems there are bugs. I hope not ;-) For example, I don't yet understand why sometimes lines inside a window-based decorations are wrapped as if there is a table that is 30% or similar of the whole window decoration width. Also there is a broken layout and uninformative error when there are consistency problems in $parent and $this on different pages. But you probably fixed this. I did not have time to make a short documentation how the new php concept works (except a mail addressed to parv and fvwm-workers a couple of weeks ago) and how new sites are added/modified on the new web tree. I will do that in a while. Just some notes here: - You need a php web server which you grants write access to the file navgen_write/navigation.inc I think you refer to the old php. Let's just suppose everyone has a new php and thus a command line php utility, so granting any write access to the web server is not needed. - After cleaning the problems reported by navgen please check your new site with the html and css checker. The problem with this is that you should first commit and only after some time (it is updated once in several hours) check out fvwm.org or fvwm.sf.net, because links like file://... or http://localhost/; can't be checked. Is there a validator that accepts input (say, as a form field) instead of url? This is not urgent and not very important at all. I plan to convert manpages and perllib documentation. Actually I am working on the creation of the manpage files at the moment. At least yesterday night ;-) The problem is that man2html produces non standard html. I found a tool http://tidy.sourceforge.net/ which can clean this html afterwards. I planed to write a new script or rewrite the actual script which makes some further adaption to get a nice manpage files for the fvwm web site. Since I might not have time till weekend you could do the manpage conversion script. Please let me know if you are going to do it so that I can send you some dones, ideas and plannings. I planned only to convert html to php, i.e. to continue with man2html. If you think it is not good enough, you may try tidy. Just, as you said the current script does dozens of small fixes of both .1 and the resulting .html, so the output becomes correct and usable. I also planned to have this structure: documentation/manpages - links to stable and unstable, and to perllib documentation/manpages/stable - contains man pages of 2.4.x documentation/manpages/unstable - contains man pages of 2.5.x documentation/perllib - links to stable and unstable, and to manpages documentation/perllib/stable - contains man pages of 2.4.x documentation/perllib/unstable - contains man pages of 2.5.x (Should we shorten documentation to doc?) Since everything is autogenerated, I see no problems to have both stable and unstable versions of man pages on the web. I will wait for your answer about whether you want to try tidy or not before starting to do this. Regards, Mikhael. -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]
Re: web pages
MG == Mikhael Goikhman [EMAIL PROTECTED] writes: MG The problem with this is that you should first commit and only MG after some time (it is updated once in several hours) check out MG fvwm.org or fvwm.sf.net, because links like file://... or MG http://localhost/; can't be checked. The fvwm.org web server pulls from CVS every ten minutes. - J -- Visit the official FVWM web page at URL:http://www.fvwm.org/. To unsubscribe from the list, send unsubscribe fvwm-workers in the body of a message to [EMAIL PROTECTED] To report problems, send mail to [EMAIL PROTECTED]