Webster's Bugs

Known Bugs

• You can not follow link that are more than 220 characters long. They should be displayed correctly, but the full value of the link is not stored so you can not follow them. (This should be fixed in 0.14 I hope)
• Changing modes reassigns the fonts, if necessary. However they will not be displayed correctly until the page is redisplayed. (i.e. reloaded or another page is displayed) Neither Will switching between Normal and MultiSync font definitions occur.
• If Fetches are done using KA9Q, there is no way of telling KA9Q to abort a fetch. (This is a problem with KA9Q rather than Webster)
• Decoding HTML is still rather slow, but getting faster. The compiled version is two or three time faster, but three or four time bigger as well!
• Formatting is not always correct
  • Single spaces may be added or ommitted, particularly just before or after a tag.
  • Tables are not completely implemented (No vertical centering for example). Also images in adjacent cells are not always aligned correctly. I hope to fix all this soon, hopefully in 0.14.
• If the program crashes, Font handles are not closed and RISC OS may run out! Useful note The leading numbers on the *FontList command are font handles, so you can close unused handles by calling 'SYS"Font_LoseFont",handle' from BASIC.
• The error Unrecoverable error in runtime system: free failed, (heap overwritten) sometimes occurs. This is from gif2spr and is due to a corrupt GIF file which the supplied version of gif2spr can not cope with. (Neither will most other RISC OS GIF decoders! Although the decoder built into Netscape usually copes, so people rarely notice the corruption)
• The FTP fetcher does not work very well as TCPIP v2.04 does not always produce prompts when commands finish. (I am told this has been reported)
• The Default External Viewers will not work until it has created the necessary files. This should now be done automatically, but you may wish to change the default actions. See the External Viewers file for details.

Reporting Bugs

If you are using the latest version before reporting the problem as a bug there are some things you should try...

• Reload the page. If this works there may be problems with the Fetcher or TCPIP that you are using.
• Make sure the error is from Webster. All Websters error boxes have Webster in the title. If an HTML page appears that contains an error code this will be from the remote HTTP server not Webster.
• Make sure TCPIP is not having problems (Search for error messages etc.). This is unlikely, but it could be having problems with your connection.
• Use the RISC OS Task Window to ensure you have not run out of memory.

When Webster gets stuck!

1. Try Holding down the ALT key on the Keyboard. In most cases this should get Webster to unwind all the loops and stop drawing the page.
2. If this works do the UnReproducable or Intermitant action below, then..
3. Try reproducing the problem by reloading the page.

For Reproducable problems

1. Turn Debuging on by changing the line in the Messages file that says DEBUG:0 to say DEBUG:2
2. Re-run Webster and reload the page.
3. Look at the Source file and work out where the decoding goes wrong. (I you can that is!) extract this piece and see if it fails on its own. You may need to extract right from the beginning, in which case the problem is not where you first though it was!
4. Mail me the information stated below.

For UnReproducable or Intermitant problems

If Webster is still running

1. If Webster appears 'stuck' start with the action above. Then
2. hold the Shift key down on the Keyboard and click with Adjust on Websters iconbar icon.
3. Mail me the result and the information stated below.

If Webster is not running.

1. Note down anything on screen and anything else you can think of!
2. If this kind of thing happens a lot, try running with Debuging turned on (See Above) Unfortunately this is slower and will corrupt the top few lines of the screen.
3. Mail me the result and the information stated below.

What to mail in bug reports

• The last 20 lines from the trace file if you could create one (Or the result of one of the actions above).
• If no images are involved, the section of the source where the failure occurs. (or all of it if you can't work out where that is)
• The URL of the page that fails. (Vital if the page needs images)
• A description of what the failure look like. (Error messages, description of whats wrong with the format etc.)
• The internal error code produced (If any)