Grok the Web

A Programmer's Guide to the New Software Development Paradigm

by Andrew Schulman

Chapter 3

Compound Documents: HTML

Last revised: April 25, 1997

• {{Chapter needs to emphasize that HTML is universal "container". Not so much this or that feature of HTML. Designers problems with HTML (see http://www.webreview.com/97/04/04/feature/index.html). But basic idea of text-based container/glue is what's key.}}
• In chapter 1, mentioned contrast between ease of compound doc in HTML vs. difficulty with MS OLE, e.g. WinWord. Some more details:

  
  {INCLUDEPICTURE "C:\\BOOKS\\controlweb\\1-5688~1.bmp" \d \* MERGEFORMAT}
  The \d if picture not included in WinWord doc file.
  Microsoft Word Err=1227 (Word cannot open the graphics file)
  On LAN, can use UNC name. Closest thing MS came to URL.
  Or if do as object:
  {LINK Paint.Picture "C:\\BOOKS\\controlweb\\1-5688~1.bmp" \a \p}
  {LINK ClassName "Filename" [PlaceReference ] [Switches ] }
  
  PlaceReference: Identifies the portion of the source file that's being 
  linked. If the source file is a Microsoft Excel workbook, the reference 
  can be a cell reference or a named range. If the source file is a Word 
  document, the reference is a bookmark.
  
  Clearly, that would be desirable in Web linking (# fragment?)
  
  \b  Inserts the linked object as a bitmap.
  \d  Graphic data isn't stored with the document, thus reducing the file size.
  \p  Inserts the linked object as a picture.
  \r  Inserts the linked object in rich-text format (RTF).
  \t  Inserts the linked object in text-only format.
  
  Can save out WinWord file as HTML; converts BMP to GIF via XXXX. 
  HTML output from Internet Assistant is crappy. Need optimizing 
  compiler for it!
  

• Object linking and embedding in Web vs. in MS OLE. OLE incomprehensible, monickers, etc., and only for one machine. On web, just works.
• OLE takes on harder problem, for sure, but burdens what do 90% of the time with what required in only 10% of cases (typical over-engineering!)
• Now, both Netscape for Windows and MSIE use OLE, so one can argue that it perhaps is a useful underlying technology. But most browsers can do <IMG SRC> without OLE. Even if genuinely useful, still have to question how many developers really should learn: put it in the browser, once, and be done with it. {{This sounds "know-nothing."}} This general theme taken up again in chapter X on web browser as "display engine."
• EXPAND!!! on the other hand, Andreesen "IMG hack" (Web Review: imghack.html): didn't go through standards, Tim B-L against autoload (Robert Cailliau in Soul of Internet, pp. 193-4??); inconsistency (why just GIF, JPEG?); compound doc draft (EMBED); also GIF legalities (CIS, Unisys). Standards process: on the one hand, IMG; on the other hand, BLICK and MARQUEE!
• Or anything with a MIME type embeddable via the <IMG SRC> tag. Yes, <IMG SRC> is limited to certain types of files. This is now a problem (see intro to RFC on compound documents), but is probably why the thing succeeded in the first place: simplicity, maybe even over-simplicity at first. This is a good opportunity to discuss Marc Andreesen's <IMG> "hack," which single-handedly caused the Web explosion.
• EXPAND!!! sidebar: IMG SRC legalities (linking vs. copying); is linking IMG like HREF, or "adding insult to injury" because stealing cycles too? (disk quotas, $), "fair use". Need separate section in chapter 5 on CGI legalities?
• http://www.intellicast.com/weather/bos/sat/

  <!--DON'T LINK TO THIS IMAGE !!!-->
  <IMG SRC="http://208.194.150.32/weather/bos/sat.wsi;alb;supsat"
  ALT="[boston satellite map]" BORDER=0></A>
  <!--DON'T LINK TO THIS IMAGE !!!-->
  

  Did reverse lookup (http://www.bankes.com/nslookup.htm): 208.194.150.32 is www3.intellicast.com; try with geturl; returns err 404
• Some sites invite linking to graphics (Map View), others like Intellicast discourage or make impossible.
• US gov: maps etc. public domain! E.g., see http://tiger.census.gov/faq.html.) (Vicinty GeoViewer license: http://yp.yahoo.com/docs/viclicense.html).
• See http://www.news.com/News/Item/0,4,9145,00.html on "Web sites feud over design"; re: copying HTML code from other sites!
• Dave Kennedy [CISSP] Research Team Chief, National Computer Security Assoc. (get URL from RMS): In an action similar to a (settled) legal threat over "inlining" of copyrighted comic strip graphics in a third party web page, a host of publishing companies have filed suit in New York City federal district court against a company called TotalNews. TotalNews uses the experimental "frames" extension to Web code to point their site's visitors to various news sources around the Web. CNN, Washington Post, Dow Jones, Times Mirror and Reuters, who have filed the suit, allege that TotalNews' practice of displaying the content of the various companies' news sites within a "frame" with TotalNews' banner ads, is a violation of the companies' rights.
• See stuff on frames/copyright in frame_copy.txt (also see http://www.eff.org/pub/EFF/Newsletters/EFFector/effect10.04)
• can have animated GIF (see http://webreview.com/books/gif/); story behind animated GIFs; did it require change to browser, or for some reason did it "just work"? (See Koman book)
• Often used in ads, such as http://www.news.com/Banners/Images/cnet.flash.banner.gif or: http://ads1.zdnet.com/adverts/imp/c00838/securewy.gif?g=r001&c=a03307&idx=3-10-10-11. Article on banner ads at http://www.msnbc.com/news/69527.asp.
• Actually, this one is GIF CGI: verified with geturl -head that this comes back with Content-type: image/gif; note interaction IMG SRC and MIME
• Multiple animated GIFs on same page: multithread?
• can have program that outputs GIF!; e.g. USNO GIF; not just object linking: process linking (see next section); GIF "on the fly"
• http://tycho.usno.navy.mil/howclock.html: uses Content-type: multipart/x-mixed-replace; <http://tycho.usno.navy.mil/cgi-bin/nph-usnoclock.gif?zone=EST&ticks=04>
• Maybe contrast Java applet at http://www.intellicast.com (can reuse by specifying full URL!):

  <applet codebase="http://www.intellicast.com/java" code="GMTClock.class" width=275 height=20>
  </applet>
  

• Another Java clock at http://www.boulder.nist.gov/timefreq/javaclck.htm (NIST Time & Frequency Division, using atomic clock)
• Good article on web time servers: http://nytsyn.com/live/Latest_columns/104_041497_094207_662.html; see http://www.as.ua.edu/math/settime.htm
• image output from program: also means can attach cameras to web, produce gif, jpg, etc. on the fly. e.g., http://www.randyville.com/cc/latest.jpg (Live view of Cuyahoga River, Cleveland OH). Or better, http://www.netscape.com/fishcam/fishcam.html (which in Netscape is CTL-ALT-F!!). Explanation at http://home.netscape.com/fishcam/fishcam_about.html. Hmm, actually this is a static GIF or JPG file that updated on disk once every minute. Following is CGI, updated every 30 sec: http://fishcam.netscape.com/fishcam/livefishcamsmall.cgi?livefishcamsmall2. Doing geturl -head on this shows "Refresh: 15" and "Content-type: image/jpeg". (Note simplicity of Refresh!! push/pull.) List of these sort of things at http://www.yahoo.com/Computers_and_Internet/Internet/Entertainment/
  Interesting_Devices_Connected_to_the_Net/Spy_Cameras/Indoor_Cameras/. A particularly interesting one is http://pilot.msu.edu/user/golebie1/fan.html (image at "http://198.109.166.172:12345/?method=push;camera=0;font=biocuse;face=b;backcolour=black;forecolour=white;position=bottommiddle;caption=LIVE VIDEO from the FAN CAM... at: %I:%M:%S%p"). Also has form to speak to him via his Mac: ACTION="http://198.109.166.172:12345", NAME="speech", METHOD="post". hidden "location"="http://pilot.msu.edu/user/golebie1/fanspeak.html". Cool NASA webcams at http://www.geocities.com/SiliconValley/1080/nasacams.html (e.g., http://163.205.10.4/shuttle/countdown/video/chan2.jpg). CUSeeMe, QuickCam. http://www.unitedspacealliance.com/live/tracker.htm is Java alternative to refresh GIFs/JPGs? Also see http://www.activeimaging.com/ (demo: Monterey Bay Aquarium; image location: http://video1.mbayaq.org/push.jpg). Squirrels at http://taycam.cc.utexas.edu/ (image at http://taycam.cc.utexas.edu/images/camera.acgi).
• multipart content-type important! (http://www.w3.org/pub/WWW/Protocols/rfc1341/7_2_Multipart.html)
• what determines what displayable with IMG SRC? MIME?
• (Note above is in cgi-bin and is obviously program because params after ?, but has .gif extension presumably to fool browser; also note nph for nonparsed header; tested with geturl -head and comes back without any MIME header: is a GIF87a file)
• perl for GIF, etc. output: see http://www.panix.com/~clay/perl/query.cgi?graphics+index (perl interface to GD; GIFgraph; graph.pm; pgperl/pgplot)
• frames make up for some deficiencies of IMG (See Webmaster in a Nutshell, chapter 4, for a nice overview of frames); saw earlier, esp. borderless frames
• Or <IFRAME> in MSIE: placed within doc like <IMG> (not supported by Netscape):

  <HTML>
  <IMG SRC="http://www.idgbooks.com/images/smallcovers/1-56884-305-4.gif"> 
  <IFRAME SRC="http://www.amazon.com/exec/obidos/ISBN=1568843054" HSPACE=3
  HEIGHT=75% WIDTH=75% FRAMEBORDER=0>
  </HTML>
  

  Scrollbars without border weird, but can see that placed within page like image.
• Can have map based on GIF: map, ismap (client-side image maps; good explanation at http://www.ee.ed.ac.uk/online/imagemaps.html: <IMG SRC USEMAP>, <MAP>, and <AREA SHAPE COORDS HREF> tags; spec by James Seidman, Spyglass)
• {{Airplane tracking based on flight number! http://www.amerwxcncpt.com/flytrax.html/ or http://tms1.vntsc.dot.gov/; http://tms1.vntsc.dot.gov/docs/asd.html is Interactive Aircraft Situation Display -- image map of US, click on dots to get more info on plane! This cool, but probably not URL related. Maybe move to area on images, compound docs, UI. See saved asd.html. Though get flight# from dot.gov, then test with flytrax (Flyte Trax Query).}}
• Another image map example: http://pubweb.parc.xerox.com/mapdocs/mapviewer.html. Xerox paper earlier.
• Image map tutorial: http://www.projectcool.com/developer/alchemy/02-imagemap.html
• Images to extend UI. INPUT TYPE=image, ability to do box around image, etc. Good example: www.lycos.com? (Maybe handle this subject in chapter 7, on web browser as display engine.)
• Good example of GIF as UI: "MSNBC Viewers Top 10" at e.g., http://www.msnbc.com/news/69527.asp (article on ad banners). Client-side image map, links to http://www.msnbc.com/modules/bom/vote.asp?points=1, http://www.msnbc.com/modules/bom/vote.asp?points=2, etc.
• Extreme example of images to extend UI: single-pixel GIF
• Putz Xerox paper also explicitly discusses incorporation of images in other docs: http://www.parc.xerox.com/istl/projects/www94/mapviewer.html#INCORPORATING: "In addition to providing an interactive user interface for browsing world maps, the map server can also provide separate map images on demand. GIF format map images can be incorporated by reference from arbitrary HTML documents... It is also possible to link the image back to the interactive Map Viewer interface so that clicking on the image will allow the user to zoom and customize the view" (this followed by an A HREF/IMG SRC example similar to my ex11).
• "A fundamental and important aspect of the design of World-Wide Web is that information references (currently implemented as URLs) can just as easily refer to dynamically generated information as static files. Although most World-Wide Web servers today perform primarily as hypertext file servers, there is an increasing trend towards more dynamic information services, where custom documents are assembled and delivered to a user on request. The presentation markup and hypertext capabilities of HTML make it possible to present dynamically generated information in a way which is both functional and aesthetically pleasing. Authored hypertext documents can be combined with HTTP based interactive services to provide documentation that is well integrated with the application's user interface. Links to appropriate documentation can appear in dynamically generated pages, and static hypertext documents can include sample links which invoke queries or other services." (http://www.parc.xerox.com/istl/projects/www94/conclusion.html). Used for example by http://www.civeng.carleton.ca/cgi-bin/quakes (near-real-time earthquake bulletin).
• One point here is how much can be done with existing HTML; surprisingly flexible. Feel we've barely scratched surface of what can be done with existing web standards, barely understand how to use, now we're off with new things (ActiveX, Java) before truly understand what already have.
• HTML & SGML purists?
• XML (http://techweb3.web.cerf.net/wire/news/mar/0318xml.html). See general description at http://www.gca.org/conf/xml/xml_what.htm. Dale says that Microsoft Document Object Model (DOM) is also relevant here. Note that MS CDF based on XML (http://www.microsoft.com/corpinfo/press/1997/mar97/cdfrpr.htm).
• Dynamic HTML: Netscape? (http://www.techweb.com/se/directlink.cgi?WIR1997030709). MS white paper: http://www.microsoft.com/workshop/prog/aplatfrm/dynhtml.htm (dynamic expansion, dynamic content: modify text; not just at load-time). http://www.news.com/News/Item/0,4,8913,00.html