Rick's b.log2 - version log

2011/11/17

• Testing with having images smaller than the cutoff size auto-reduced by 50%, so text will reflow sort-of logically around insert images on mobile devices. I'll play around, see what I think of the layouts...

2011/11/06

• Fixed up some missing stuff in the mobile view. Corrected a few bugs.
• Added detection for... Android, Blackberry, Opera Mini/Mobi, iPad/iPhone/iPod, Palm OS, Symbian, Windows Mobile, Kindle, and also Mobile, Pocket, Smartphone to catch some others.
  You can use keitai to force behaviour; but note that keitai=0 from a mobile device is currently buggy (links will probably revert you to mobile mode). This is because every link must be altered to notice if "keitai" is specified, and to pass it on. It's either that or use cookies, and you know how much I love cookies...
• Image scaling tweaked to 308 pixels, which is a 320px viewport with 6px margin either side (for scrollbars, etc). This works well on my Android phone (354px, as it was, did not!).
• I believe my phone can do a 480px viewport (more in landscape mode), however it seems from my investigation that 320px is the de facto best choice, without trying to locate and read mobile RDF descriptions on the fly for custom formatting. 320px looks good, so 320px it will stay for now.
• All of this folded into the main blog code.

2011/10/27 - dev

• Now uses mobile selector to choose which resource set to load. This gives flexibility to modify the layout specifically for mobile devices - I suspect I might try to auto-generate a simple calendar 'image' to replace the table...
• Ditto for the CSS.
• No longer auto-provides the last entry if no date is specified. It will have the same behaviour, but it will do so by forcing a hard-redirect to the content with a date. This should make it clearer that undated entries should not be linked to or indexed in search engines.

2011/10/26

• Began work on support for a mobile version. The idea is instead of having a sprawling layout with side bars and such, to be a lot more restrained. Here's a screenshot of Amazon's mobile site, which is quite pleasing to use on a mobile:
  
• The mobile version does not, as yet, detect mobile devices. You can force it by setting the parameter "keitai" to be '1'. Keitai, for what it is worth, is Japanese for "mobile". ☺
  Here's a link to make it easier: http://www.heyrick.co.uk/blog/?keitai=1
• At this time, all images larger than 354 pixels width will be scaled down to be 354 pixels.
• All images will be converted to JPEG with a 60% quality setting. This means the visual quality will be less impressive, but it'll be quick to download, especially on slower (EDGE, WAP, etc) networks.

2011/02/01

• No longer records User Agent strings. There doesn't seem much point when awstats does it better...
• Added a little QR code easter egg.

2011/01/16

• Some internal tidying of code and rationalisation for alternative (dev) versions.
• Added browser/OS popularity links. This is not as returned by awstats, there's a script that fetches the page and processes it before display (mostly to botch the graphics links out of the original HTTPS domain as my site does not have a security certificate and your browser would complain, a lot).

2010/12/19

• GeoLoc retrieval method altered because new host blocks fopen() accessing external sites (good security measure, that!).

2010/12/11

• Numerous changes to aid program/logic flow and internal tidy-ups.
• Caching control relaxed for entries by given date.

2010/04/07

• Following a number of printouts I have which hold neither the URL nor the date, plus since moving away from IE, the URL is no longer 'embedded' in the saved HTML file, I have modified my b.log code so just underneath the copyright message is the complete URL of the document, plus when you retrieved it.
  If you are using a modern standards-compliant browser, you will not see this as you browse the site. It is only intended to be visible on printouts...

2010/04/04

• Spent much of today doing a big rewrite. Lots of changes. It is still a largely monolithic piece of code, but it is better than it was before. And there have been some changes outwardly visible...
• Reorganisation of the right-hand panel. Promoted most recent entries, and demoted search (hardly ever used :-( ), and added HTML4.01/CSS validation flags.
  Note many pages will NOT validate 100% due to deftitle and deftag; I could retrofit comment markers around it for every page, but there's little point as it makes not one iota of difference to the browsers. So I'll call it HTML 4.01 (loose) compliant. If anybody argues...
• Podcast/audio links are given as text ("audio available") in the calendar pop-ups, and as little speaker icons in the summary list.
• You can now ask for a summary of every entry written.
• Numerous tweaks to the style sheets, for full CSS validation. The validator doesn't seem to like some named colours ("cyan") while being quite happy with others ("yellow"), so now it is all hex values and God help us if this needs to be changed later on. Nothing says darkgreen quite like #006400, don't you think?
• Style sheet expanded for a special printout mode. When printing, you want the goodies without the crap. Especially the crap designed purely for navigation which serves no purpose in a printout, and has an ink-consuming background.
  • The title is rendered as black text on white (except the 日本語 one which is an image, hit refresh...).
  • The ID thing (upper right: you are w.x.y.z, pleased to meet you!) is omitted.
  • The entire right panel is omitted. All of it.
  • The comment explanation and the comment write-a-comment form are omitted; but comments themselves remain.

2010/04/01

• Fixed quirk that made the search form not work. Not that anybody actually ever uses it... :-(

2010/03/27

• Various small changes to pass HTML 4.01 loose (non-frameset) validation; plus a positional change for a quirk in Opera (img "absbottom" is different in Opera than IE8 and Firefox).
• Added "last five entries" to the right-side panel.

2010/03/25

• Various small changes to the layout and styles.
  I wanted to make this document HTML 4.01 strict compliant to work in standards mode instead of the current Quirks mode (which is not entirely the same on all browsers); but this means fully adopting CSS.
  While I am not against this in principle, I am not happy with the way the site would then render on non-CSS capable browsers. There is no "font", "bgcolor" and such. It is all handled quite nicely by CSS. As for incapable browsers... um... welcome to HTML 2.0!
  Thus, if it looks slightly different on the various browsers, so be it. I'm aiming for widest support, not best compliance.
• There is some news, however. I will no longer be testing any part of my site using IE.
  Why? Well...
  • IE6 has finally been laid to rest, though numerous companies are still stuck with it as certain custom software only works specifically with IE6. It isn't just a situation that IE is the only browser to handle ActiveX objects, but also IE6 has so many standards quirks that it is not always possible to update.
  • Certain governments advise against using IE. Some say dump IE6, some say dump IE.
  • IE8 on my machine takes several times longer to do anything than my habitual browser, Firefox. The only thing IE8 does quickly is load, but given the integration between OS and IE, it is hardly a fair comparison.
  • Try installing IE6 and IE8 on the same machine and switching between them. Oh, wait, you can't...
  • Given that XP is a huge market share of internet use, even today, it seems to be incredible that Microsoft announce that IE9 will use special Direct2D GPU-assisted graphics acceleration. And because of this, it'll not work on anything pre-Vista. Like XP. Gun meet foot...
  • Microsoft claimed that they want to embrace standards, like HTML5 etc, but they wish to downplay the significance of the Acid3 test. This is probably because IE8 scores a truly pathetic 20%. Current builds of IE9 are reported to be around the 50% mark. Other browsers are typically in the mid-to-high 90s or a full 100%.
  I will recant my decision if somebody can give me a reason to support IE. And I mean, a better reason than "I still use it!".

2010/03/21

• If a region is not available, GeoLoc will no longer say:
    b.log last read at 12:34 on 2010/03/21 by somebody in  (France).
  It will now say:
    b.log last read at 12:34 on 2010/03/21 by somebody in France.

2010/03/20

• Added a little "easter egg" where every so often (it is random), the "Rick's b.log text is displayed using various Japanese characters, like this:
  

2010/03/14

• Search now linked to site look'n'feel.
  Matches now show formatted date (yyyy/mm/dd).
  The title description is also now shown.
  The excerpt has the search term highlighted in bold uppercase.
• Search is still NOT sorted - I'll need to read all the file names into an array, sort, and read back out. Later.
  Search can be 'funny' if you do daft things, like search for 'a'.
  Searching for punctuation generally doesn't work. This is not so much a bug as an implementation issue.
  Searching searches the entries data, so will not include user-added comments. This is not a bug.

2010/03/13

• Added on-screen trap for people who try to add extra parameters. There ARE no extra parameters, don't waste time looking (though, to be honest, it looks like 'bots throwing a list of known-vulns at detected php files).
• Pre-alpha of the search script. It is grotty, it is not at all integrated with the b.log look'n'feel (I began coding it an hour ago...), but it may be useful to people?
  Just don't expect too much - it reads entries, strips HTML, looks to see if there are any matches for the supplied content. Oh, and it is very literal so if you search for "the the the" then you'll see pretty much every page listed three times.
  That said, the methodology of the search is unlikely to change much now; if you want fancy you should Google. This is just a way to look for phrases without "guestimates".
• IMPORTANT! The search does NOT (yet? ever?) support quotes. You cannot look for risc os as you'll get every 'os' word listed, like dross or cosmos. Search for something more specific (in this case, RISC) and use the mini-preview text to pick the useful entries.

2010/03/08

• This log switched to list most recent entries first, so new stuff will be at the top where you'll see it easily.
• Added a CYAN marker in the calendar for spoken entries.
  The "current" entry will always be yellow, but there will also by cyan for audio-assisted ones.
• Updated the older MP3 files with the b.log2 URLs.

2010/03/08

• Now supports the NiftyPlayer for MP3 podcasts.
  You can see this in action here and here.
  More podcast versions to come...
• Fixed the auto-converted older entries so they didn't monumentally screw up <pre> formatted text.
• Added code to attempt to block my site appearing in other people's frames. This might affect Google/Yahoo image searching, but it is mainly for people who create framed sites and don't bother validating their off-site links to ensure it correctly discards the frames (otherwise my entire site would load into the current frame!).
• Checked through all the older entries to ensure they look correct.

2010/03/06

• Potential vulnerability closed. It may have been possible to inject PHP scripting into various support files which would then be executed as the page is created.
  I say "may" because the file type used is not one that Apache understands as containing script, so the scripting was ignored (but on my off-line copy and on the live server); however that is no justification for a potential gotcha... consider the door slammed in its face.
• The "you're a bot" message moved to a separate file. Easier than a heap of echo statements!
• Separation line added after each comment.

2010/02/21

• Comment system changed to support the use of 'real' newlines.
• Comment code now uses server name, so no more screwups like local server sending comment tests to the live server. :-)
• Local server now moved to /blog path in prep. for autoconversion of older entries.
  Just gotta write some code to deal with that...

• Older entries converted (largely by software). Every one has a comment section now (original comments imported from older b.log).
• Calendar clipping set to June 2008, as this is when my b.log was started.
• RSS link added to the top.

2010/02/20

• The no-date-show-last default behaviour is disabled if the visitor is determined to be a 'bot. This is because I only want valid links to be indexed, as the show-last behaviour means the "last" will keep changing.
  Likewise the GeoLoc is not updated for 'bots.
  A 'bot is determined as any host that contains "bot" or "crawl" or "spider" in the name.
• The old b.log link now has the nofollow attribute.
• The comment entry box now has a reminder of \n and \p for newline/para.
• Comment handler now accepts \\ for a \ character.
• Removed "background.png" from the page wrapper, it is no longer relevant... and no longer there, it's in /images anyway. ☺

2010/02/19

• Invalid date help message now includes an example.
• Added link back to the old b.log.

2010/02/18

• Tweaked GeoLoc to not bother looking up 127.0.0.1 when running on my local server!
• Stricter validation on date - if you added extra digits to the end it would not be faulted and would instead return the latest entry. Now it will correctly whinge.

2010/02/17

• Making "/summary not /00" behaviour broke the month displayed in calendar, fixed.
• Added GeoLocation to the right-hand menu.
• Various minor layout tweaks.
• Added itty-bitty help text just below the calendar.
• Fixed broken commenting - was linked to old comment handler on /blog - duh!

2010/02/15

• Current entry date highlighted in yellow.
• Date ranges now validated. [thanks, Rob]
• Month summaries now display "yyyy/mm/summary" instead of day "00".
• Link to next entry at the bottom, if there's a next to link to. [suggested by Rob]
• Link to return to the top of the page at the bottom.
• Style tweaks so IE8 and Firefox look alike. [IE5 and IE6 don't, they're broken...]

2010/02/14

• Initial version