Differences

This shows you the differences between two versions of the page.

versiontwentytwocleanupdone [2007/03/02 13:45] (current)
Line 1: Line 1:
 +====== 0.22 DONE ======
 +
 +This is what has been Done on 0.22, taken from VersionTwentyTwoCleanup.  This is just a temporary archive to make sure nothing gets overlooked or any bugs reappear.  Also a few things listed here are not going into 0.22 but may be added later or at least thought about later so we don't want to loose them.
 +
 +  * bds: Single Message View shows date using the html_date_format instead of the long date
 +  * UI Advanced page - location of configuration file is missing
 +  * the old default skin (SimplyBlue) disappeared (for no apparent reason) when the skins were converted to the new format
 +  * XMLRPC.pm is not compatible with the new template-based UI so it cannot be configured via the UI
 +
 +  * jgc: test that Refresh is working on the History page
 +  * jgc: Why isn't history entry cleanup happening?
 +  * jgc: figure out what's happening with set_query when a search is set
 +  * Manni: those logger > errors are because you haven't update the log_() calls in IMAP to have a level
 +  * History page uses MM/DD/YY format for the date - will this format be made confiugurable? (I'd like DD/MM/YY)
 +    * Me, I like DD.MM.YY.
 +    * jgc: Take a look at Localize_Date in English.msg
 +  * What is the call to $self->db%%__%%()->quote( $hash ); doing in history.pm? If IMAP.pm cannot do the same thing, I'm lost jgc: that's for database safety
 +  * Who put all the tabs in history.pm?
 +  * Trying to adapt the IMAP module, I sometimes get this when calling history::get_slot_fields():
 +<code>DBD::SQLite::db selectrow_array failed: near "c13557464d30452b78a96b14bac": syntax error at /POPFile/History.pm line 356. </code>
 +JGC: You are calling get_slot_fields with a slot and not a hash right?
 +MH: No. Sorry.
 +  * Following console messages appear when starting up:
 +
 +
 +<code>   Initializing...
 +        {core: history mq windowsUse of uninitialized value in concatenation (.) or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 363.</code>
 +Use of uninitialized value in numeric gt (>) at C:\PROGRA~1\POPFILE/POPFile/Logger.pm line 197.
 +Use of uninitialized value in concatenation (.) or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 363.
 +Use of uninitialized value in numeric gt (>) at C:\PROGRA~1\POPFILE/POPFile/Logger.pm line 197.
 +<code>config logger}</code>
 +POPFile Engine v0.21.1 running
 +
 +Use of uninitialized value in concatenation (.) or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 130.
 +Use of uninitialized value in concatenation (.) or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 130.
 +
 +JGC: Manni, fix the log_() calls to include a level and the Logger ones will go away.  I've fixed the MQ problem.
 +  * When number of messages per page is set to 20, up to 19 messages are shown per page (setting 21 per page shows up to 20 per page). The "Recent Messages" count shows 102 but when page size is set to 20 there are 5 pages with 19 messages and one page with 1 (a total of 96 msgs). With page size set to 21 there are 4 pages with 20 msgs and 1 page with 17 (a total of 97 msgs).
 +  * Advanced page shows history_history_days = 2 but on Configuration page the "Number of days to keep" box is empty (not sure if this was done deliberately since history can't be deleted at the moment)
 +  * Still getting a couple of console messages when starting POPFile:
 +
 +
 +<code>   Initializing...
 +        {core: history mq windowsUse of uninitialized value in numeric gt (>) at C:\PROGRA~1\DEMO/POPFile/Logger.pm line 197.</code>
 +Use of uninitialized value in numeric gt (>) at C:\PROGRA~1\DEMO/POPFile/Logger.pm line 197.
 +<code>config logger}
 +        {interface: html}
 +
 +        {classifier: wordmangle bayes}
 +        {proxy: pop3}</code>
 +
 +(This is a "clean" POP3 install on Windows - no IMAP.pm, no XMLRPC.pm)
 +
 +
 +  * Bayes.pm begins logging before Logger.pm has setup it's correct path, resulting in two log files if the logger_logdir is set to anything other than $POPFILE_USER (this same problem is in V 21, see bug 967966)
 +  * Parsing the header fields to get the hash:
 +    * I suggest to get rid of any \r and \n before the real parse.
 +    * Include the most recent received line.
 +    * Here's some code that seems to work quite well and that could replace what is currently done in commit_history%%__%%:
 +<code>   my %header;
 +   my $last;
 +
 +
 +   if ( open FILE, "<$file" ) {
 +       while ( <FILE> ) {
 +           s/[\r\n]//g;
 +
 +           if ( /^$/ ) {
 +               last;
 +           }
 +
 +           if ( /^([^ \t]+):[ \t]*(.*)$/ ) {
 +               $last = lc $1;
 +               push @{$header{$last}}, $2;
 +
 +              
 +           }
 +           else {
 +               if ( defined $last ) {
 +                   ${$header{$last}}[$#{$header{$last}}] .= $_;
 +               }
 +           }
 +       }
 +
 +       close FILE;
 +   }        
 +   my $hash = $self->get_message_hash( ${$header{'message-id'}}[0],
 +                                           ${$header{'date'}}[0],
 +                                           ${$header{'subject'}}[0],
 +                                           $${$header{'received'}}[0] );</code>
 +  * jgc: Localized date/time format for footer's current data/time display
 +  * jgc: Add back some notion in the History of each "session" with a horizontal line
 +  * Classification Accuracy "graph" missing from Buckets page jgc: it's time had come, it caused too much confusion for too little benefit so I nixed it.
 +  * I get strange sorting on the history page when sorting by "From".
 +  * The new history divider does not show up (used the 'default' skin but could not see anything)
 +  * jgc: Set default sort in History to -inserted
 +    *  Will Date::Parse help? I tried this yesterday with ppm to no avail. Had to install TimeDate. (Dunno, I'm on Linux and had to install Date::Parse (and only Date::Parse) to get the cvs version to run - Scott)
 +    * To get it working on Windows all I had to do was install TimeDate and copy a few things to minimal Perl
 +  * jgc: Dynamic column resizing in History page
 +  * ISO-encoded subjects do show up in encoded form.
 +  * tomvoss: Magnets page: The links to individual numbered pages do not include the session key. The links to next and previous do.
 +
 +  * jgc: design a better magnet icon
 +  * Some messages in history have no "To" listed but Single Message View shows there is a "To" line in header (current (10 June) CVS has fixed this problem)
 +  * Rebuild popfile.exe, popfileb.exe, popfilef.exe, popfileib.exe and popfileif.exe so they can be used to run the latest CVS version.
 +  * jgc: Implement Delete Checked
 +  * Can we make the logger give us the contents of @{$self->{last_ten%%__%%}} if we get a level 0 message and logger_level is set to 0?  This is a question please ask it in the forum
 +    * jac: did anything happen on log last ten?
 +  * logger.pm contains tabs.
 +  * Logging in tabbed format yields \t instead of tab: need double-quotes on line 240.
 +  * The popfile.exe dated 10 June does not work (it does not appear to call any of the four EXE files)
 +  * jac: remove directory sleetimages from CVS, when getting CVS version it appears as a skin (+ lavishimages).
 +    * jac: I just downloaded from CVS to a new directory and an empty lavishimages appeared.
 +  * jac: Merge patch to add session seperator to windows skin.
 +  * jac: Footer is inside main PF table on Buckets page.  See windows or default skin for good example.
 +  * jgc: make magnets display correctly in the History
 +  * jgc: Implement Delete All and Check Page
 +  * jac: If you change the order of the //html_columns// in Advanced, it changes only the column name in history, the data in the columns remains in the original order. jgc: don't do that!  -- jac: why not?  Confusing PF is fun.
 +  * jac: When using a short date format //%m/%d// and sorting on the Arrived and possibly Date column the Sort indicator > or < may get pushed up above the column name.  There should be a non breaking space (nbsp) between the indicator and column name.
 +  * jac: Default columns.  I think by default we don't need to show the CC column.  The default skin has a large font setting so takes up a lot of space.  We don't want a new user's first PF experience to be figuring out how to make the columns fit.
 +  * bds: The messages\history_cache file was left behind after message history was converted
 +  * Messages with missing subjects get no clickable link. We need a <no subject> subject for those.
 +  * jac: Configuration page - change "<current log file>" to "<View current log file>" for more clarity to new users.
 +  * jac: Future enhancement: unclassified bucket is hard to see on skins with dark background (blue skin).  Either unclassified color should be configurable or skins should be able to define a color set.  I think it would be easy if you could seperated color definition into a file in each skin directory.  Inside PF you could access colors by index number and the actual color and color name would be specified by the skin.  With encouragement that skin designers stick to the pallet as much as possible. jgc: sounds like an RFE.  -- jac: I bet there is a closed RFE or three for this.  I will submit a new one eventually.
 +  * jac: Arrived and Date display should be configurable to not display time and or year.
 +    * bds: The time and date format can be changed using the html_date_format on the Advanced page. For example "%d %h %H:%M" displays "12 Jun 09:48". To find out more, read the documentation for Date::Format
 +    * jac: Remember PF is aimed at regular users, most people aren't going to care about the exact time or be comfortable editing html_date_format on the Advanced page.
 +  * Reclassify doesn't work.
 +  * jac: Could we not display the SOCKS V prox * stuff unless a use Socks config option is turned on.  Most people are never going too need it.
 +  * jac: Magnets page should change "Value" text box to "Values" since it now accpets multiple values for magnets.
 +  * jac: Didn't you say you were going to add a few other magnet options. jgc: not in this release
 +  * jac: "Recent Messages (21[Remove All])" is pretty ugly.  Who cares how many recent messages and the Remove all is ambiguious and dangerous at the top like that.  Does it remove all messages or those 21 recent messages?  Something that removes more than one message should be at the bottom.
 +    * bds: "Recent messages (21)" means there are a total of 21 in the history (it saves you having to work out the total yourself).
 +    * jac: Its still ugly and not very useful.  And I really don't think the Remove All button should be there.  If the number has to stay I can live with it.  I know its fixable by editing the skin, but we need to come up with the best overall default skin we can.  And then document any additional variables that skinners may access.
 +    * jgc: this is a discussion, take it to a forum
 +  * jac: think about deprecating some of the older skins that don't look quite so good: default (very dull color scheme), orange (orangecream is very similar and much better).  orangecream, simplyblue, coolblue, or glassblue would make a good default skin setting.  - No deprecation, people use those skins.
 +  * jac: skins lavish, sleet, and sleet-rtl need help with too much centering - I will do that if you can't find the original skinners.
 +  * jac: "[Remove checked]" button should also be at the top (below search), we used to have PF like this with it only at the bottom, but after several complaints (from people with extremely long history pages) you added it at the top too.  jgc: there's no space for it at the top, like there is for Reclassify. - jac: I didn't say there was room, just said people are going to complain about it.
 +  * jac:  History page: "Change width of ..." seems odd where it is.  It is easy to use there, but could be on Config page in "History Columns:" next to each column checkbox.  That would be less automatic, but would allow user to customize much more.  That may be over doing the customization.
 +  * jac: Magnet creation message when creating multiple magnets needs a newline between.  Currently the different messages run together.
 +  * bds: Reclassify does not work from Single Message View
 +  * jac: include magnet graphic/icon [[http://sourceforge.net/tracker/index.php?func=detail&aid=973498&group_id=63137&atid=502958| See patch]].
 +  *bds: Message arrival times are wrong. Messages appear in the wrong order when History page uses the default sort order (or when it is explicitly sorted by Arrival time). The session separator no longer works properly. For example two messages received 90 minutes apart (message folder files timestamped 12:03:22 and 13:35:16) are treated as being in the same session (Arrival times 10:10:55 and 10:10:59 respectively)
 +    * MH: I don't see any of this. What is wrong about the sort order?
 +    * bds: [[http://sourceforge.net/forum/message.php?msg_id=2622918| Further details in forum]]
 +  * bds: Default logger_level 0 does not provide much information in the popfile log
 +    * MH: That's not a problem, that's a feature. What information are you missing?
 +    * bds: Default log setting no longer shows, for example, when messages are removed from history.
 +    * MH: The plan was to have a log that shows only errors for the lowest level.
 +    * jac: Is level 0 going to be default?  I think we want something closer to what we have now by default.  For support we often ask people what is in their PF.log.  If its not an actual "error" PF logs we won't be able to help until they turn up the logging level and the problem happens again.  For errors that are infrequent (like PF freezing on Windows) we would not have any useful information to troubleshoot.  Too little in the logs may just frusturate users in need of help.
 +       * MH: That's why I suggested to keep a buffer of the last ten (or so) log messages and output that buffer in case an error is logged.
 +    * bds: Reducing the number of messages in the log makes the dozens of spurious "Discarded" messages look more important than they are
 +  * The sort_xy headers that are used for sorting history entries aren't decoded before they get into the db. I think we need to call decode_string() on them too.
 +  * jridge:  Messages received that contain the "Subject:" header, but no text following this cannot be clicked on in the history UI because nothing is displayed under the Subject column.  Note that jgc already fixed the similar issue when the "Subject:" header is missing from the message entirely, now POPFile displays "<subject header missing>" in the UI for these cases.
 +  * Warnings during update of the old database:  DBD::SQLite::db do failed: PRIMARY KEY must be unique at /Classifier/Bayes.pm line 650, <INSERT> line 35.
 +  * bds: Some messages are being correctly classified and passed to mail client but they do not appear anywhere in the history or in the message folders. This is on Windows with forking enabled. Problem seems to occur when downloading from more than one account at a time.
 +  *bds: History conversion creates bad entries (i.e. 'committed' not set to 1) in history table for messages whose CLS files refer to non-existent buckets
 +  * bds: The X-POPFile-Link displays the history page instead of Single Message View
 +  * bds: popfile.exe generates 41 spurious "Discarded unknown parameter" messages in the log file every time it is run (still happens with current CVS (19 June) and popfile.exe dated 11 June)
 +  * bds: Downloaded messages from more than one account on Windows at the same time (with forking enabled). All appeared in history but the message counts on the bucket page were lower than expected - it seems only the data from one of the accounts was used to update the totals. This is repeatable (with UI displayed during download).
 +  * bds: On the History page, Search From/Subject appears to be operating as Search From/To instead
 +  * bds: Although the History page (and Recent Messages count) is updated while mail is downloaded (a manual refresh is needed to see the new data), the Classification Accuracy and Messages Classified per bucket counts are not updated on the Buckets page. POPFile has to be shutdown and restarted to get the Buckets page to show the new data. If POPFile is started and the UI is _not_ displayed then the data for the Buckets page is updated as messages are downloaded but once the Buckets page is displayed, the data on it stops being updated. (Manual refresh and restarting the browser have no effect - POPFile needs to be restarted to see the new statistics)
 +  * bds: At startup get two "uninitialized" messages:
 +Initializing...
 +<code>        {core: history mq windowsUse of uninitialized value in join or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 373.</code>
 +Use of uninitialized value in join or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 373.
 +
 +and another two when startup is completed:
 +
 +
 +POPFile Engine v0.21.1 running
 +Use of uninitialized value in join or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 126.
 +Use of uninitialized value in join or string at C:\PROGRA~1\POPFILE/POPFile/MQ.pm line 126.
 +  * bds: Current CVS (23 June, 15:25 GMT) crashes when Configuration page is accessed. The first error message was HTML::Template->new() : Cannot open included file C:\PROGRA~1\POPFILE/skins/default/Platform::MSWin32=HASH(0x68d2f4) : file not found. at C:/Program Files/POPFile/lib/HTML/Template.pm line 1580 (followed by other template errors)
 +  * jgc: fix up SMTP, NNTP, and MSWin32 modules with templates for their UI components
 +  * bds: 'Show word frequencies' and 'Show word probabilities' crash POPFile (same messages in each case): POPFile Engine v0.21.1 running
 +Use of uninitialized value in hash element at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2208.
 +Use of uninitialized value in hash element at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2209.
 +Use of uninitialized value in subtraction (-) at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2210.
 +Use of uninitialized value in subtraction (-) at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2210.
 +Illegal division by zero at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2210.
 +  * Selecting Single Message View from history page shows the message with only "<Previous" and "Close" navigation shown. Clicking "<Previous" results in the display of an empty message (no From:, To: , Subject: or classification data, no body) with "Next>" and "Close" navigation and some bucket/count/probability data (from who knows where). Current CVS no longer has navigation links in Single Message View so the problem has gone away
 +  * bds: Enabling html_test_language makes the UI almost unusable (eg buttons have identical text)
 +    * jac: not only is it unreadable, the buttons and other UI elements don't work since the HTML code to build them is truncated by extra quotes.
 +  * bds: The new 'Show word scores and decision chart' generates lots of console messages (all the same): Use of uninitialized value in subtraction (-) at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2194. (The chart appears)
 +  * bds: For some messages the decision chart does not have any bars shown. Examining the web page shows bar width is less than 0.88 - IE6, Firefox 0.9 and Mozilla 1.7 all had trouble (IE6 sometimes manages to show one bucket's bars but not always the same bucket).
 +  * bds: 'Show word frequencies' and 'Show word probabilities' generate the follwoing messages (same messages in each case):
 +<code>Use of uninitialized value in hash element at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2208.
 +Use of uninitialized value in hash element at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2209.
 +Use of uninitialized value in numeric gt (>) at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2210.
 +Use of uninitialized value in numeric gt (>) at C:\PROGRA~1\POPFILE/Classifier/Bayes.pm line 2210.</code>
 +  * jac: Configuration page Module Options has some extra "SOCKS V proxy *".  Possibly we are just missing a header explaining that the second set goes with another module. jgc: it's part of POP3 -- jac: I see two textboxes for "SOCKS V proxy host:" and "SOCKS V proxy port:" for a total of 4 SOCKS boxes.  Are there really supposed to be 4?  Another option to lessen the number of SOCKS boxes would be to enter the host and port as host:port into one box.  Anyone setting up something like this should be familiar with that common format.
 +
 +    * MH: Do you still see this with the current CVS? I guess that these belonged to nntp or smtp.
 +    * jac: Seems to be fixed now using Brian's Installer test version.  I still think SOCKS support should be enabled or disabled so the extra text boxes don't confuse people that don't even know what SOCKS is.  The toggle to display the settings could even be right there in that pannel. -- now I realize it is an optional install so I guess it wouldn't show up normally anyway.
 +  * bds: On Configuration page the Yes/No buttons for "console window" and "tray icon" do not use the "toggleOff" class
 +    * jac: Was this related to Manni's [[http://sourceforge.net/tracker/index.php?func=detail&aid=918617&group_id=63137&atid=502958| patch]]? If so, can it be closed?
 +  * jac: Updated all skins for sesion seperator (for more info see Release Notes section).  Will submit patch.
 +  * bds: When SMTP module is enabled the UI locks up when the Configuration tab is selected (SMTP template file not found)
 +    * mh: Either I forgot to commit a file (smtp.pm it seems) or cvs is currently having problems. I can't access developer CVS right now and anonymous cvs gives me the same broken combination of smtp.pm and the templates. It's noted however and will be fixed soon (if sf plays along).
 +      * mh: Seems I really forgot to commit smtp.pm. At least the popfile-commit archives don't show a change in that file. Very good time for a cvs blackout.
 +  * mh: history::is_valid_slot() will feed anything to the db without any prior parsing. This can result in lots of warnings and a broken single message view if that function was called from html::url_handler that was passed an illegal jump_to_message URL.
 +  * jac: In search box entering an address as //"Allandlocators" <allandlocators@earthlink.net>// causes trouble.  PF needs to replace the &lt; and &gt; for the search, currently they break the HTML tag.  This also happens in 0.21 so its not a new bug.
 +    * mh: what kind of trouble? what html tag? Do you get the desired search results?
 +    * jac: The search doesn't work, the text box has only the part in quotes (with the quotes, which should have been good enough for finding the message) but didn't because of the broken tag I guess, the text box tag I think closed early.
 +      * mh: OK, there is a rendering issue that could be solved by replacing " with &amp;quot;. But the search nevertheless works for me. I don't think we need to replace < and > although that wouldn't hurt.
 +      * jac: the problem I am seeing is //" />// displaying on the rendered page after the text box.  Looking at the html code makes it seem that the problem is the > closing the tag early.  This is in Mozilla, Opera may behave different since this is an odd situation.
 +  * Single message view now has problems with emails containing lots of addresses in the to or cc fields: The reclassify control is invisible in most browsers with most skins. I suggest to change the head of the single message view to a single table (instead of two) so that from ... date, bucket, should be, etc all end up on the left and take a complete "line".
 +    * bds: Is this still a problem? I had a message with 52 entries listed under "To" and Single Message View was OK
 +      * mh: Yes, still a big time problem. It may depend on the skin, but I tested a couple of skins and always got the same (broken) result. If nobody thinks that we must have a two column layout on the header of the single message view, I will fix this.
 +      * mh: Found out that skins are not the issue, but browsers are. Opera and IE will render the two tables side by side, no matter whether there is enough space or not. Mozilla et al will show the second below the first.
 +  * jac: How about put the magnet icon in reclassify column?  We have the room and its not being used in Magnet case, plus it looks strange breaking up the line up of bucket names with the image. -- Also it looks funny when wrapped [[http://www.quarter-horse.com/popfile/lavishmagnet.gif| Lavish screenshot]] -- this is fixed in my skins update.
 +  * jac: Top Reclassify and Remove buttons appear to be bolder (header class?) on Windows skin on Konqeror.  The upper Reclassify button is inside a <th>, the lower one is inside a <td>.  How do we solve that?
 +  * jac: Config page "Delete Now" and its checkbox should have  between them.
 +  * jac: Apply buttons on Config page's User Interface panel need space before button, Same on Filter By on history page.
 +  * jac: Need space between UI and logger dropboxes and their buttons on Config page.
 +  * jac: XMLRPC on config page's Apply button should be on line with input box.
 +  * jac: Advanced page Update and Bucket page Apply buttons do not have a CSS class, shouldn't it have //class="submit"//, you can see the problem on StrawberryRose skin, its not noticable on others.  -- Done in my skins update.
 +  * jac: On Magnets tab I think we can remove the ":", its inconsistant with our other text entry configure interfaces.
 +  * jac: the magnets display table rows should stretch across the whole body so it matches the history rows.
 +    * mh: why should the magnets table match the history table? If the rows in the magnets table hold little information that fits on X% of the screen, why stretch it?
 +  * jac: Config page Choose skin dropdown is smaller than it should be, it is most noticable on prjstealbeach and blue.
 +    * mh: I still don't get how large it should be. The width is determined by the width of the content unless a skin defines a minimum width for drop-downs.
 +    * jac: It is just a difference of a couple pixles in width and height.  Its not a very visible problem.
 +  * jac: "Remove Checked" doesn't fit on button using Sleet skin.  I tried to contact Dan, hopefully he can fix it.
 +  * jac: its not in CVS yet, but I have an ok fix for this.  It looses the rounded corners on buttons with long text, but it works.
 +    * jac: if we get a fix from Dan for this fine, if not its good enough.
 +  * jac: How about a Gmail "snippit" like column option?  Gmail shows the first line of emails and its often useful to distinguish spam from nonspam there, but their spam detection sucks compared to POPFile.  The first line of the email body could be stored in the DB which would also also searches on at least that tiny piece of the body.
 +    * jac: not important, maybe next version.
 +  * jac: Mark noticed the "filter setting is remembered between sessions, the negate is not" -- Sam fixed this.
 +  * Default SMTP and NNTP to Off (they currently automatically try to start if found in Proxy at startup)
 +    * Not quite done, yet. SMTP and NNTP default to off, but proxy.pm overrides this.
 +  * jac: Where did the "Remove Page" button go? We cannot get rid of that. It is useful if you aren't ready to clear all history but perfer to remove things manually. Also this option sent removed messages to the Archive, Remove Checked doesn't.
 +  * jac: Think about going back to old style color chooser. Opera does not show colors in the dropdown box, only the color names so choosing a color in Opera is not very nice. Discussion on this is in my color.def thread [1]. (color.def should wait till after 0.22)
 +  * Massa: make the windows version consume much less memory; v0.21 consumes about 25MB here, that is much too much! The high memory consume is the reason which detains a lot of people from using it on older machines.
 +    * jac: Not much we can do, most of the memory requirements are from Perl and SQLite which are required for POPFile to work.  Related [[http://sourceforge.net/tracker/index.php?func=detail&aid=987160&group_id=63137&atid=502959| RFE]].
 +  * bds: popfile.exe generates 41 spurious "Discarded unknown parameter" messages in the log file every time it is run (still happens with current CVS (23 June, 15:25 GMT) and popfile.exe dated 11 June)
 +  * jac: "(Change width of From, To, Cc and Subject columns: ...)", is ugly and needs help. What about a "Adjust columns" link that sets a condition that will allow those resize buttons to show? Then normally only the one "Adjust columns" link would show, when active the adjustment buttons show.
 +  * jac: From the Advanced page the From, To, CC, and Subject columns should be resizable seperatly. The normal Column Resize option would then just add or subtract 1 from each column width as needed. This would allow more user customization.
 +  * jac: The accuracy bar is not working. If it was removed on purpose I can see why, it just caused confusion with some people and didn't really mean anything. Just checking.
 +  * jac: Not sure if its repeatable, but if there is only one page of history the page and Refresh link don't appear at the top of the page sometimes. I have only see this once and it was just after converting, before I had a history page 2. Once I got a page 2 and deleted enough messages to get back to only 1 page the link appeared.
 +  * jac: When removing all messages from a page you are sent to a blank history page, we should go to the previous page in history and display actual messages I think.
 +  * jac: insert.sql is leftover after the upgrade. Depending on your corpus size it can be a very big file. Is this file kept on purpose, is it a temporary file for the upgrade?
 +  * jac: UI for setting date format on history page? Possibly just dropdown of a few predefined options.
 +  * jac: I get 2 "Use of uninitialized value in string ne at /UI/HTML.pm line 876." messages in the console on Linux when viewing the Configuration page.
 +  * bds: When the message history gets converted the old history_cache file is not deleted
 +  * jac: On Linux settings changes on the Config page are not written to popfile.cfg imediatly. When changing a setting PF give this console message, "Use of uninitialized value in string ne at /UI/HTML.pm line 877." PF does write changes to the cfg imediatly when hitting update on the Advanced page. If this isn't due to the bug, why not write it when a change on the config page is made? Crash or shutdown improperly and you loose any recent settings. I got a bunch of these console errors and I am not sure all were from accessing the Config page (HTML.pm version 1.282):
 +<code> Use of uninitialized value in string ne at /UI/HTML.pm line 877.
 + Use of uninitialized value in string ne at /UI/HTML.pm line 877, <$reader> line *.</code>
 +  * jac: Still on Linux, now with HTML.pm v1.283 I still get the line 877 error and now 2493 with a line number at the end that starts at 1 and increments (1, 2, 3, 6, 8, 9 ...):
 +<code> Use of uninitialized value in string ne at /UI/HTML.pm line 877.
 + Use of uninitialized value in quotemeta at /UI/HTML.pm line 2493, <MESSAGE> line *.</code>
 +  * bds: Latest CVS (with all the script file changes) still generates dozens of "Discarded unknown parameter" messages in the log when popfile.exe is used to start POPFile. With the basic POP3 configuration 43 of these messages are added to the log _every_ time popfile.exe is used. [My reports on this problem keep getting moved to "Done" yet the problem still occurs]
 +    * jac: Done is pretty general, it also seems to mean "fix later" or "not worth holding off release".
 +  * bds: Current versions of the popfile*.exe files (dated 10 or 11 June) do not use the new Loader.pm (1.21)
 +  * bds: The UI templates do not work as expected. Instead of simply putting the modified template file in the skin folder, it may be necessary to copy all of the template files even if only one is being changed. A suggested fix is in the [[http://sourceforge.net/forum/message.php?msg_id=2691936| SourceCode forum]]  (HTML.pm 1.295 fixes this problem)
 +  * bds: If the number of recent messages is not an exact multiple of the number of messages per page then the last page of the history will not appear (clicking "Next" on the penultimate page or clicking the number of the last page simply re-displays the penultimate page - but the browser address changes as if the last page is being displayed). No console error messages appear.
 +  * jac: Running PF on Linux I consistantly get a broken pipe when I double click on a link or click very quickly on two different links.
 +  * bds: Latest CVS (with all the script file changes) still leaves behind an empty "pdk-*" folder in the TEMP folder _every_ time popfile.exe is used
 +  * jac: Decide if Trayicon should be disabled by default since it appears to be a problem more and more frequently.
 +  * Windows service PF should be able to run as LocalSystem, 0.21 version must run under a normal User.
 +  * bds: The logger_logdir parameter appears to have no effect - the log files are always created in the POPFILE_USER directory.
 +  * jac: Delete empty message sub-directories after all the messages are expired from a directory.
 +  * bds: When upgrading the message history, the following console message appears for every message which does not belong to a bucket: Use of uninitialized value in addition (+) at C:\PROGRAM FILES\UPGRADE/Classifier/Bayes.pm line 328.
 +
 +
 +====== UI ======
 +  * jac: History column resizing should have a lower limit. Currently it allows 0 length for the columns. Then when you decrease it again it goes to 39 (automatic setting?). If you decrement down to 0 you can have 0 character columns, if you try to set 0 on the Advanced page it seems to give you Automatic. Setting the lower limit to 10 should be enough, even at that setting there is little useful information given so its still probably far lower than people will ever need even on a PDA.
 +  * jac: Why are you allowed to turn off the bucket column in history? Subject probably doesn't need to be turned off either? The UI is pretty useless without the subject. Less options under History Columns would look better too.
 +  * jac: Date on Detailed message view page should not depend on date setting, just print date header from the email. With my date setting all I have is "Sun 13:55". That is good for History page, but full date would be more useful on the "DETAILED" view.
 +    * mh: I kind of agree. But the Date should still be formatted according to the user's preferences. But since the pretty_date function doesn't discriminate single message view and history page, I don't think it's worth the bother.
 +  * jac: Should html_cache_templates be tuned on for speeding up UI when its released? Does this option do anything?
 +    * mh: Nothing that would be noticable for me.
 +
 +====== Release Notes ======
 +  * Note new Perl Module requirements in changelog and documentation for v 22:
 +    *Date::Parse
 +    *HTML::Template
 +    *HTML::Tagset *
 +    *DBI (could this be more specific or do we need entire thing?) *
 +      *bds: The Windows installer only installs lib\DBI.pm and lib\auto\DBI\DBI.* from the DBI package
 +    *DBD::SQLite * (I don't think current docs on Linux install mention this)
 +    * bds: ActivePerl? needs the TimeDate? module, not just Date::Parse (i.e. it needs Date::Format, Date::Parse and Time::Zone)
 
versiontwentytwocleanupdone.txt · Last modified: 2007/03/02 13:45 (external edit)

Should you find anything in the documentation that is incomplete, unclear, outdated or just plain wrong, please let us know and leave a note in the Documentation Forum.

Recent changes RSS feed Donate Driven by DokuWiki
The content of this wiki is protected by the GNU Fee Documentation License