NewsFox
| Home Installation Source Code Members Bugs Screenshots Forum |
Version 1.0.2 has a new NewsFox function: filtered web pages. This allows you to download web content to examine while offline. But it allows you to filter the web content before it is stored, both to make it smaller and give it a higher signal/noise ratio.
Here is how it works. A new column is added to the article pane(labeled '+' in en-US). It is hidden by default, so you'll have to enable it with the columnpicker(column named 'Filtered Web' in en-US). The icon for this column has five states:
- read icon(small gray dot): use the usual description from the feed as before
- blue downarrow: the filtered web page is being downloaded
- unread icon(larger green dot): the filtered web page will be displayed
- unread icon with yellow background: images from the filtered web page are being downloaded(the filtered web page will still appear when selected)
- red X: there has been an error, make sure
javascript.options.showInConsoleis equal to true, rerun if necessary and then look at the error console to see errors
newsfox.x.removeXbody. If this value equals 1(default), the filtered web page will be discarded. If this value equals -1, the filtered web page will be retained. If this value equals 0, the user will be asked whether to delete the filtered web page.
Now how do you tell NewsFox how to filter the web page? In the feed options dialog, there is a new tab 'Web filter' that lets you construct a filter. This is quite an advanced feature, and problems with its use will typically be outside the scope of NewsFox technical support. There are two methods you can use: 1) regular expressions and 2) JavaScript. You do either by entering the filter inside the text box. You can also just augment the usual feed description by checking the box to download images and not adding a filter.
Regular Expression method: You enter a regular expression and the text/html from the article link page matching the regular expression becomes the filtered web page. The default uses the regular expression <p.*?<\/p> which selects all the <p> elements from the web page and throws everything else away(assuming the <p> elements are closed). To construct a more useful regular expression, you need to look at the text/html from web pages you are interested in and find useful markers to use.
Here are a couple example general references on regular expressions:
JavaScript method: You enter JavaScript in the box. The last value will be used as the filtered web page. This must evaluate to a string. The JavaScript is evaluated in a sandbox. The following variables are available: linkDOM - the DOM of the article link page, linkHTML - the text/html of the article link page, win - a copy of the window object of the iframe the result will be displayed in(changes to win do nothing, only the last value of the JS makes any difference), doc - a copy of the document object of the iframe the result will be displayed in, getElementsByClass(class, tag, node) which returns an array of elements with given class and tag contained in node, getDomAndHtml(site) which returns an object with properties HTML and DOM giving the text/html and DOM of the given site. getDomAndHtml(site) should be avoided as it needs to work synchronously and thus locks up the entire browser while the request is being processed. An example of JavaScript to enter in the box to return the entire web page would be
getElementsByClass("","html",linkDOM)[0].innerHTML;
or
linkHTML;
Note that external CSS is not downloaded in either method. JavaScript on the page is not run, which may make some aspects of the page unavailable. linkHTML and linkDOM are just the text/html and DOM of the given page with nothing else loaded.
R Pruitt (wa84it AT gmail.com)
This beta release will become version 1.0.2 after translations and bug fixes.
The usual disclaimers apply: this is a beta release so use it with caution on a backup of your Newsfox folder. I use it on my real newsfox folder.
The new features (where to look for bugs to fix):
- Filtered web pages
(bug#19171)
Read bug#19171 and separate post for more information. - Selection after deletion improved
(bug#19419)
The selection now goes to the next undeleted article rather than the next article. If there is no next undeleted article, the selection goes to the last undeleted article. If there is no last undeleted article(ie all articles are deleted), the selection does not exist. - CSS: newunread property
(thanks K!r!lleXXI)
Articles freshly downloaded are given the CSS property 'newunread'. This is removed upon either another refresh of the feed or closing NewsFox. There is no default styling for this property - if you want to use it you need to put a line in newsfox.css inside your profile folder. Example:treechildren::-moz-tree-cell(newunread) { background-color: #eee; }Note that if you change the font-weight, you need to be careful of the order of the 'unread' and 'newunread' properties in the CSS files. - Separate article deletion confirmation (bug#19200)
- cs-CZ locale added (thanks Yearling at Babelzilla)
- Minor
- SeaMonkey locale switching bug
SeaMonkey doesn't switch locales in extensions. I'm not a SeaMonkey user other than seeing NewsFox installs and runs, so I've never known this. NewsFox has a workaround now. - Parent for link checked in RSS feeds
I've never seen this in the wild, but it came up with an invalid feed mentioned in the forums. - newsfox.advanced.aggregate
If you have a slow computer, or never look at feeds in groups, you can set this equal to false and see a speed improvement when groups are selected. - newsfox.advanced.backupFeedFiles
(bug#19280)
Setting equal to true will backup each of the feed files, doubling the size of your NewsFox folder. There is now also an attempt to recover from writing errors even if this is not set. - More invalid feeds handled
Several feeds set their overall namespace as xmlns="http://backend.userland.com/rss2", thereby making the feed invalid. NewsFox now has a workaround for these.
- SeaMonkey locale switching bug
NewsFox 1.0.1 is released.
- keywords
- keyboard shortcuts editing in options dialog
- more properties available for skinning in tree
- source in view-pane clickable (and link works now for atom feeds)
- newsfox.advanced.flagTagged
- media:content support
- Minor
- ties in article titles sorted correctly
- interest filter improvements(more garbage words ignored, probs updated after delayed scoring)
- textview.css is now a chrome file imported with local textview.css
- newsfox.global.confirmDeleteArticle controls article deletion confirmation (bug#19200)
- multiple article selection doesn't display article (bug#19223)
- little dialogs mostly to yes/no rather than ok/cancel where appropriate
- option menu improved
- some tooltips, menuitems changed to reduce translation duplication(357 strings -> 297 strings, then of course some new)
- groups not collapsed until return OK in group options dialog
- search folder icons in feed options dialog
- if newsfox.z.dragFeedOutOfGroupIsCopy=false, dragging feeds out of group is now move, ctrl-drag to copy
Keywords can be added in the options dialog or in newsfox.keywordX in about:config where X=1,2,3,.... More than 5 keywords can be added in about:config. Articles with keywords in the article body have green text in the article pane, and the keywords are highlighted in green in the view-pane. Technically, rows in the article tree with keywordX have properties keywordX and keyword. The default styling in newsfox.css is
treechildren::-moz-tree-cell-text(keyword)
{ color: #10a010; }
treechildren::-moz-tree-cell-text(keyword,focus,selected)
{ color: white; }
Keywords in the view-pane are all of class keyword and the default styling in textview.css is
.keyword
{ color: green; font-weight: bold; }
So, it is possible to change the article-pane styling to be different for each keyword, but all keywords are treated the same in the view-pane.
I didn't realize this was an essential feature until I added it. I have a few search groups that just search on a single term, for example I have one that searches 'newsfox' so I don't miss anything someone says about it. It is much better to have a keyword. Keywords are only matched in article bodies and are case insensitive. There can be spaces, but they must match exactly as well. Technically: [articlebody.toLowerCase().indexOf(keyword) > -1] is a match.
If newsfox.advanced.flagTagged is equal to true, articles are flagged whenever they are tagged. Then they won't be deleted automatically by accident.
See the beta release notes for more information.
R Pruitt (wa84it AT gmail.com)
The feed pane and article pane of NewsFox are XUL trees. Trees can be given properties that can then be used in CSS to give styling to the trees. The trees have row properties, column properties and cell properties that apply to row, columns and cells. Cells also have the properties of their row and column. This is important since certain aspects can only be styled via cells rather than rows or columns. For instance text properties only come from the cell. This article lists the properties available for these trees in NewsFox. Properties listed on a single line are mutually exclusive.
Feed Tree
- Row properties
- folder, feed, category
- feedsFolder, tagFolder, searchFolder
- feedsFeed
- unread
- Cell properties
- faviconcol
A row has the property feedsFolder if it is the FEEDS folder, that is, it is the first row.
A row has the property feedsFeed if it is a feed within the FEEDS folder.
A row has the property unread if there are more than zero unread articles associated with that row.
A cell has the property faviconcol if space needs to be left for an icon, that is, if it is not a category.
Article Tree
- Row properties
- noID
- flagged
- unread
- highprob
- prob1, prob2, ..., prob10
- hour1, hour3, hour6, hour12, hour24, hourother
- day1, day7, day30, dayother
- keyword
- keywordX where X=1,2,3,4,...
- Column properties
- flag, title, read, date, author, source, blog, prob
- Cell properties
- faviconcol
A row has the property noID if NewsFox can't construct an ID for the article. If an article has no ID element (
Scaled probabilities from the interest filter are rescaled to fill the interval from 0 to 1 based on the group/feed you are looking at. So even if all the absolute probabilities are near zero, the scaled probabilities will still be spread from 0 to 1.
A row has the property highprob if the scaled probability is greater than 0.98.
A row has the property probX if the scaled probability is in the Xth decile(10%) from the top. An article with a scaled probability of 0.82 is in the 2nd decile from the top and hence has property prob2. Note that articles with scaled probabilities higher than 0.98 will have both properties highprob and prob1.
A row has the exactly one of the hour properties. If the age of the article is less than X hours, the article will have property hourX where X=1,3,6,12,24 and X is as small as possible. If the article is more than 24 hours old it will have property hourother.
A row has the exactly one of the day properties. If the age of the article is less than X days, the article will have property dayX where X=1,7,30 and X is as small as possible. If the article is more than 30 days old it will have property dayother.
A row containing a keyword has the property keyword.
A row containing keywordX has the property keywordX where X=1,2,3,4,5,....
A cell has the property faviconcol if space needs to be left for an icon.
Examples
Base
Here are the relevant parts of the built-in newsfox.css file:
treechildren::-moz-tree-cell-text
{ padding-left: 4px; }
treechildren::-moz-tree-image(faviconcol)
{ height: 16px; width: 16px; }
treechildren::-moz-tree-cell-text(unread)
{ font-weight: bold; }
treechildren::-moz-tree-cell-text(noID)
{ text-decoration: line-through; }
treechildren::-moz-tree-cell-text(highprob,unread)
{ color: #500000; }
treechildren::-moz-tree-cell-text(highprob,unread,focus,selected)
{ color: white; }
treechildren::-moz-tree-cell-text(keyword)
{ color: #10a010; }
treechildren::-moz-tree-cell-text(keyword,focus,selected)
{ color: white; }
treechildren::-moz-tree-row(current)
{ border: 1px dotted Highlight; }
treechildren::-moz-tree-row(current,selected)
{ border: 0px; }
Notice that text modifications (color, font-weight for example) are on the cell-text not on the row. Firefox also has some built-in properties(e.g. current, selected), but I don't have references with me on the road so you'll have to search for them if interested.
Feed Tree
Here is an example for the feed tree:
@-moz-document url("chrome://newsfox/content/newsfox.xul")
{
.feedtree > treechildren::-moz-tree-row(feed)
{ background-color: lightgray !important; }
.feedtree > treechildren::-moz-tree-row(feedsFeed)
{ background-color: red !important; }
.feedtree > treechildren::-moz-tree-cell-text(unread)
{ font-style: italic; }
.feedtree > treechildren::-moz-tree-cell-text(searchFolder)
{ color: blue !important; }
}
Try and keep that FEEDS folder closed by making the feeds hard to look at when it is open! Note that the unread cells still have font-weight: bold since we didn't turn it off.
Article Tree
Here is an example for the article tree:
@-moz-document url("chrome://newsfox/content/newsfox.xul")
{
.articletree > treechildren::-moz-tree-cell-text(noID)
{ color: white !important; }
.articletree > treechildren::-moz-tree-column(flag)
{ background-color: blue !important; }
.articletree > treechildren::-moz-tree-cell(flagged,flag)
{ background-color: red !important; }
.articletree > treechildren::-moz-tree-cell-text(prob1)
{ color: #900 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob2)
{ color: #800 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob3)
{ color: #700 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob4)
{ color: #600 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob5)
{ color: #500 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob6)
{ color: #400 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob7)
{ color: #300 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob8)
{ color: #200 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob9)
{ color: #100 !important; }
.articletree > treechildren::-moz-tree-cell-text(prob10)
{ color: #000 !important; }
.articletree > treechildren::-moz-tree-cell-text(highprob)
{ color: #f00 !important; }
.articletree > treechildren::-moz-tree-cell-text(hour1,date)
{ font-size: 100% !important; }
.articletree > treechildren::-moz-tree-cell-text(hour3,date)
{ font-size: 95% !important; }
.articletree > treechildren::-moz-tree-cell-text(hour6,date)
{ font-size: 90% !important; }
.articletree > treechildren::-moz-tree-cell-text(hour12,date)
{ font-size: 85% !important; }
.articletree > treechildren::-moz-tree-cell-text(hour24,date)
{ font-size: 80% !important; }
.articletree > treechildren::-moz-tree-cell-text(hourother,date)
{ font-size: 70% !important; }
}
Make titles unreadable when deleted?! Change just the flag column background to blue, unless the article is flagged then make it red. Make higher probability articles have redder text. Note the position of highprob. Make dates(just the date column) shrink as the article ages. Article title displays can be styled differently for each keyword, rather than the same styling for each keyword as is done in the base example.
This beta release will become version 1.0.1 after translations and bug fixes.
The usual disclaimers apply: this is a beta release so use it with caution on a backup of your Newsfox folder. I use it on my real newsfox folder.
The new features (where to look for bugs to fix):
- Keywords
Keywords can be added in the options dialog or in newsfox.keywordX in about:config where X=1,2,3,.... More than 5 keywords can be added in about:config. Articles with keywords in the article body have green text in the article pane, and the keywords are highlighted in green in the view-pane. Technically, rows in the article tree with keywordX have properties keywordX and keyword. The default styling in newsfox.css istreechildren::-moz-tree-cell-text(keyword) { color: #10a010; } treechildren::-moz-tree-cell-text(keyword,focus,selected) { color: white; }Keywords in the view-pane are all of class keyword and the default styling in textview.css is.keyword { color: green; font-weight: bold; }So, it is possible to change the article-pane styling to be different for each keyword, but all keywords are treated the same in the view-pane.
I didn't realize this was an essential feature until I added it. I have a few search groups that just search on a single term, for example I have one that searches 'newsfox' so I don't miss anything someone says about it. It is much better to have a keyword. Keywords are only matched in article bodies and are case insensitive. There can be spaces, but they must match exactly as well. Technically: [articlebody.toLowerCase().indexOf(keyword) > -1] is a match. - Keyboard shortcuts editing in UI
As far as I know, there is no good way to tell whether a key combination will work other than by trial and error. - More skinning properties
Enough to write a separate article. As an example, the following should be unobjectionable, but noticeable so that users may get interested:treechildren::-moz-tree-cell-text(highprob,unread) { color: #500000; } treechildren::-moz-tree-cell-text(highprob,unread,focus,selected) { color: white; }The property 'highprob' is added to articles with a scaled probability of greater than 0.98 in the interest filter. - Source in view-pane fixed
The source is now a link, and the link should work correctly in atom feeds. - newsfox.advanced.flagTagged
If this is equal to true, articles are flagged whenever they are tagged. Then they won't be deleted automatically by accident. Default is still false. Only available in about:config. - Media:content support
Why can't people just use <enclosure>? - Minor
- Print article added to options menu
- Interest filter improvements
More garbage words thrown out, html tags, etc. Probabilities are saved to disk after delayed article scoring. - Default textview.css imported
There is now a textview.css file inside NewsFox so no textview.css is created by default in the newsfox folder. The program textview.css is loaded as styles and then any textview.css in the newsfox folder is loaded. This should give much the same behavior as before, but is much cleaner. Now I don't have to worry if there is a .keyword declaration for instance, I just put it in the program textview.css. - Mini-dialogs to yes/no where appropriate
Of course, maybe people are so used to seeing ok/cancel when the answers should be yes/no that they'll be thrown by being given the correct choices. - Improved options menu
The unneeded of 'group options' and 'feed options' are not shown. Recover simplified: I don't think the try first option was particularly useful. - Translation strings reorganized
- If newsfox.z.dragFeedOutOfGroupIsCopy=false, drag is now move, ctrl-drag is copy
Firefox has problems with the choice popup anyhow, and this is standard OS behavior. The default is true, so this is advanced anyhow.
Since version 0.7, the file textview.css in the newsfox folder in your profile folder has allowed customization of the view-pane of NewsFox. The feature was improved in version 0.8.1.
Since version 0.8.2, image files placed in the images subfolder of the newsfox folder have been able to replace skin image files. With version 1.0, it is also possible to replace mail.png and encl-*.png as well.
In version 1.0, the file newsfox.css in the newsfox folder is a global stylesheet. In particular, it will apply to NewsFox. Previously, some people have used Stylish(Firefox extension) to style NewsFox. It is now much easier. This post will discuss a few of the things you can do with this new feature.
This feature can be accessed in K-Meleon by pointing your browser to chrome://newsfox/content/newsfox-overlay.xul which will load the stylesheet and then open NewsFox.
The following discussion is aimed at users who know a little CSS - just like me - so it will be the blind leading the blind. Resources: an introduction to CSS, access to the newsfox.xul file. Choose view: text near the top of the page and you should be able to look at the most recent newsfox.xul. You can also get the file from the extension you installed if you know how. You need to be able to look at the newsfox.xul file to find the ids and classes in the file.
Here is part of an example file just to illustrate a few points.
@-moz-document url("chrome://newsfox/content/newsfox.xul") {
#tBcheck, #fBcheck /* to use default Firefox icon */
{
list-style-image: url("chrome://browser/skin/Toolbar-small.png") !important;
-moz-image-region: rect(0px 64px 16px 48px);
}
#fBtag { visibility: collapse; }
#feedTreeChildren:-moz-tree-twisty(closed)
{
list-style-image: url("data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAkAAAAJCAYAAADgkQYQAAAATklEQVR42mP8%2F%2F8%2Fw6NHjxhwAXl5eQZGkCJXV9f%2FQP7RuXPn2uBV1NHRwVBRUXEUJIGsGEMRDCArxqkISTHDnj17GEk3iaCbCPqOUDgBAKwXUgoTciZ1AAAAAElFTkSuQmCC") !important;
}
#newsfox-panes,
#mainToolbar,
#busyAnimation
{ -moz-box-direction: reverse; }
treechildren::-moz-tree-cell-text(unread,title)
{ font-stretch: narrower; }
treechildren::-moz-tree-cell-text(noID)
{ text-decoration: none !important; color: red !important; }
#newsfox-window,
splitter,
treecol,
.treecol-image
{ background-color: #abc !important; }
}
Since the file is a global CSS file it is a good idea to restrict the scope of your CSS with @-moz-document, note the trailing '}' closing the opening brace on the first line. Here I have removed the '!important' that aren't necessary, but you can put '!important' on every declaration.
The first group gets the image for the main toolbar refresh icon(#tBcheck) and feed toolbar refresh icon(#fBcheck) from the usual place that Firefox looks for its refresh icon. So this may not work on SeaMonkey, for example. It definitely doesn't work on K-Meleon as they don't have chrome icons in the sense that Firefox does.
The next hides the tag icon from the feed toolbar.
You can specify an icon inside the newsfox.css file. This might be useful if you wanted to give your newsfox.css to someone else without having to worry about exchanging icon files. You can turn an icon file on your computer into a data: URL here for example.
You can reverse the order of the NewsFox panes.
The next makes bold text take up a similar amount of room as regular typeface. The property 'title' is the id of the title column, so this only applies to the title text. Firefox automatically makes the id of columns a property that can be used in CSS.
The next shows how to use red text for deleted articles. You need to turn off the strike-thru with 'text-decoration: none'.
The last shows how to color the background. I don't know how to color the scrollbars.
R Pruitt (wa84it AT gmail.com)NewsFox 1.0 is released at mozdev. Addons.mozilla.org(AMO) will not allow extensions to declare Firefox versions of 3.0.* until FF3RC1 is released, so this won't be released there until then. At least, that was the original plan. With all the big late changes (bugs 424923, 292789, 429785 for example) to Firefox3, I am not nearly as confident as I once was that this version will work with FF3. It works at least with the 080418 nightly on Windows XP :).
- newsfox.css in newsfox folder is a global CSS file, in particular, it works on NewsFox
- keyboard shortcuts help improved
- canceling OPML import from 'start new' fixed, thanks jackbin
- accel.xml file now UTF-8 and well-encoded, thanks KirilleXXI
- partial fix for bug#18860(see bug)
- future date set to current time when requested (bug#18866)
- Minor
- defaults for new users: catSort=1, noneStrict=false, invalidStrict=false
- XMLhttprequest fixed for Firefox3 no longer allowing cross-site requests
- article link not explicit in textview
- mail.png and encl-*.png are now grabbed from newsfox/images/* if available
- new icons
- bad feeds display troubleshoot when selected
- ctrl-alt-uparrow prints textview pane
The file newsfox.css inside the newsfox folder inside your profile folder is now a global CSS file for Firefox. So you can skin NewsFox. There is a customization forum. Also see the feed post about newsfox.css. Since this is a global file, it is a good idea to format your newsfox.css file this way:
@-moz-document url("chrome://newsfox/content/newsfox.xul")
{
/* CSS goes here */
}
The keyboard shortcuts window is non-modal, that is it can stay open while you are using NewsFox. It can also be dragged by grabbing and dragging the window, not just the titlebar(or whatever your OS allows).
By popular request, showing invalid and non dates will not be done by default. Added benefit for developers of no user complaints of feed dates not being interpreted correctly by NewsFox.
Firefox disallowing cross-site XMLhttp is good for NewsFox. Error reporting is better: you actually get to see the error code if there is a server problem. Feeds that have permanently redirected URIs are updated automatically in the feed options. Not updating permanent redirects and updating temporary redirects may be an option in the future.
The article link is no longer explicitly displayed. Firefox changed its handling of whitespace, so the link was wrapping, and then I realized there was no reason for having it. The link is now given both in the title and the weblink icon. Hovering over either will display the link text in the status bar. To regain the old view of the title (it will still be a link) add the first line
.newsfox-title { text-decoration: none; color: black; }
.newsfox-link { display: none; }
to your textview.css file. The second line keeps the weblink icon from displaying if that is desired.
Technical note for my own recollection: to get mail.png and encl-*.png from local files, they have to be translated to data: URIs since the text-view pane has no access to file:// or chrome:// URIs. So now the original icons have to be translated to data: URIs as well!
Bad feeds don't display 'troubleshoot' if they are already selected when they become bad, since you might be reading something else.
Printing will probably not make it into the UI. Firefox3 has print this frame available from the standard context menu, so it should be possible to find. NewsFox has no access to the text-view pane, all clicks and keyboard events there are handled by the OS, and no javascript is allowed. This is at the direction of the Firefox masters on pain of .... So a print button would have to be outside the text-view pane. Access might be put on the options menu.
R Pruitt (wa84it AT gmail.com)This beta release will become version 1.0 after translations and bug fixes.
The usual disclaimers apply: this is a beta release so use it with caution on a backup of your Newsfox folder. There is not much content change from 0.8.4.4 - I use it on my real newsfox folder.
The new features (where to look for bugs to fix):
- New keyboard shortcuts help pane
NewsFox 0.8.4.4 is released. This minor release will only be on mozdev.org. For those of you wondering, 0.8.4.3 is in bug#18709.
- rel="self" links in atom feeds incorrectly processed (bug#18752)
- Default tooltip not displayed on long article title
- No autosubscribe warning only given once (bug#18709)
- 5 zh-TW strings updated
Links with rel="self" that would be incorrectly processed are rare in the wild.
R Pruitt (wa84it AT gmail.com)NewsFox 0.8.4.2 is released. This release will only be on mozdev.org as it is minor.
- Firefox 3 implementation of live bookmarks
NewsFox 0.8.4.1 is released. This release will only be on mozdev.org as it is minor, and some of the new hotkeys may change depending on feedback.
- Hotkeys for view pane (bug#18635)
- All text view html attempted in xhtml first
- Feed loading speed improved
- Tooltip hack for FF3.0 removed
- Minor fixes
- Memory leak fixed (bug#18627)
- Articles displayed from top always
- Toolbar hiding
The new hotkeys for the view pane can be viewed here. Ctrl-shift-left-arrow replaces select-prev-word, but it should be relatively easy to get the behavior with select-prev-char and autorepeat. Comments/criticisms welcome on the bug. The list will eventually be translated and available inside NewsFox.
Text view html is attempted in xhtml first, so that SVG (for example) will display.
Firefox bug#413268 is fixed, so the hack for article tooltips is removed. This means version 0.8.4.1 article tooltips will not work in FF3.0a1-FF3.0b3, but will work with FF2 or current builds.
Hiding of the toolbars is available on context menu (names will get translated in next translated version).
R Pruitt (wa84it AT gmail.com)NewsFox 0.8.4 is released.
- New column in article pane: Pr (interest)
- OPML import speed improved
- New columns in article pane: source, author, feed
- Option to delete files and preferences when uninstalling
- Bookmark partial syncing (see bug#14980)
- Most preferences now available in user interface (bug#18149)
- Alternate category sorting (bug#18305)
- hu-HU Hungarian locale (fenya25, Babelzilla.org)
- Minor fixes
- Global about:config option to delete old news (bug#18454)
- Workaround for Firefox grabbing article when column sorted (FF bug#413266)
- 'New tag' available on context menu or on add group menu
- More bad feeds accepted
- Control-shift-a selects all articles in article pane
The new interest column(label: Pr) is a Bayesian filter to indicate your probable interest in an article. The ordering should be paid attention to, rather than the actual values. It uses the same statistical method as in spam filters. This is more difficult as your interests are probably broader than a spammer's interests. This starts with no base information, you need to read and download a bunch of articles before it will work well. It only updates probabilities when a feed is refreshed. My default sort order is now interest-descending followed by read-descending.
New article columns source, author, and feed are all hidden by default. Use the columnpicker to select them.
Bookmark partial syncing is provided. To enable it, check the appropriate option on the 'general' tab of the global options dialog. When disabled, NewsFox will work as before. When enabled, a special NewsFox bookmark is created: title is 'NewsFox sync' and the keyword contains a list of feeds and update times. NewsFox does the following when a feed is refreshed:
- The special bookmark is retrieved and the feed update time is set to the newer of the existing update time and the update time in the bookmark(if any). Articles are downloaded, but only articles newer than the new update time are marked unread.
- If feeds exist in the bookmark, but not in the current NewsFox, the user is given the chance to add these feeds.
- An updated version of the special bookmark is written to bookmarks.
Alternate category sorting is provided. The about:config option is newsfox.advanced.catSort with options 0: alphabetical(current), 1: alphabetical with unread articles followed by alphabetical with no unread articles, 2: decreasing number of unread articles. Option 1 makes feeds with many categories easier to use.
R Pruitt (wa84it AT gmail.com)This beta release will become version 0.8.4 after translations and bug fixes.
The usual disclaimers apply: this is a beta release so use it with caution on a backup of your Newsfox folder. I use it on my real newsfox folder.
The new features (where to look for bugs to fix):
- New column in article pane: Pr(interest)
This column is a Bayesian filter to indicate your probable interest in an article. The ordering should be paid attention to, rather than the actual values. It uses the same statistical method as in spam filters. This is more difficult as your interests are probably broader than spammer's interests. It is an ongoing project. There are 10 parameters to the model - they probably won't be in the UI, but write me if you would like to have them in about:config prefs. I might consider putting them there. Examples of parameters: decay rate for old info, # of occurrences of word before using in score, # of words to compare, prior probability of looking at unread article, maximum # of words to keep, when to throw words away. This starts with no base information, you need to read and download a bunch of articles before it will work well, if it ever does. It only updates probabilities when a feed is refreshed.
I will change my default sort to interest-descending followed by read-descending. - New columns in article pane: source, author, feed
These are all hidden by default. To display them, use the column picker at the upper right corner of the article pane. Things to check: does author column work for all feeds, do articles sort properly. - Option to delete files and preferences when uninstalling
You probably won't be inclined to test this. Known bug: can't change newsfox directory alert comes up if NewsFox open when deletion happens(still works though). - Bookmark partial syncing
This may disappear in future versions depending on the development of Weave and online storage solutions.
To enable this, check the appropriate option on the 'general' tab of the global options dialog. When disabled, NewsFox will work as before.
When enabled, a special NewsFox bookmark is created: title is 'NewsFox sync' and the keyword contains a list of feeds and update times. NewsFox does the following when a feed is refreshed:- The special bookmark is retrieved and the feed update time is set to the newer of the existing update time and the update time in the bookmark(if any). Articles are downloaded, but only articles newer than the new update time are marked unread.
- If feeds exist in the bookmark, but not in the current NewsFox, the user is given the chance to add these feeds.
- An updated version of the special bookmark is written to bookmarks.
Hopefully this will be outmoded soon. Based on my experience with Foxmarks during testing, Foxmarks(with the Foxmarks server) is not reliable enough to use for important data. - Most preferences now available in UI
Available from the global options dialog. - New tag available from context menu and new group button
- Alternate category sorting
About:config option is newsfox.advanced.catSort with options 0: alphabetical(current), 1: alphabetical with unread articles followed by alphabetical with no unread articles, 2: decreasing number of unread articles. Option 1 will become the default in the next version. This is so current users won't have behavior changed without their input. - Small parser changes
More bad dates and links in feeds accepted. - Control-shift-a selects all articles in article pane
I don't know how to get an XUL tree to use control-a (trees seem to eat this combination).