9.0201

[MikeGale]

I got around to running tests on 9.0201.

Results follow:

1) This was on a Win 2003 system, with the pro version.

2) I tested with the batch tool.

3) First test of several (maybe 6) batches, probably more than 200 pages, gave what I expected.

4) On exiting I got a swarm (maybe 5 or 6) exception type messages. I sent the first, one killed the rest.

5) I attempted to create an exception again. First one batch then shut, then two batches... three batches... four. In none of these did I get an exception.

6) Some messages, out of windows, suggested that CSE had run foul of the data execution protection mechanism. On checking DEP I wasn't certain that I had the relevant executables protected. (The naming displayed was generic, I couldn't guarantee it was 9.0201 or even version 9 I had listed there.) I added the exe and the command line processor.

Current status. Uncertain. I'll run and see whether I get exceptions. (Current best guess is that it is a DEP issue.)

[Albert Wiersch]

Thanks Mike!

I'm about to release v9.0210 which may improve this as I believe I've found and fixed an issue that could cause the accessing of invalid memory. You might want to wait for that update before running more tests.

[Albert Wiersch]

I just made available v9.0211 for registered std/pro users as well as a trial version and a lite edition. Whew!

[MikeGale]

Hi Albert,

Thanks for that. Sounds like a lot of work.

I just tested, much the same as before. Batch processor several batches a few hundred pages.

All went as expected, and when I shut it down no messages. I think you may have hit the source of that issue.

I did get one new issue.

It is reported as

'An error occurred trying to run the JavaScript linter. Error: 2005052001: can't open outputfile "...\Temp\~cse_jsmsg43323888.tmp"'

I got much the same message twice. I haven't noticed it before.

[Albert Wiersch]

All went as expected, and when I shut it down no messages. I think you may have hit the source of that issue.

That's good!

I got much the same message twice. I haven't noticed it before.

Hmmmm... can you easily reproduce this or give me more details?

[MikeGale]

Hi,

I ran two tests (one with a minimal batchfile, the other the same batch file used yesterday). The message did not re-appear.

I then checked the temporary file alluded to in the previous message.

That file was not there, but a similarly named one was. (Last digits of name differed.) This file, from it's time, was created during my tests today.

This file contained:

C:\_ScriptLibrary\***.js(0:0):can't open file: No such file or directory

0 error(s), 0 warning(s)

Which I think might nail this. (Maybe I get a visible log when that temp file can't be created. I still don't know why creation wasn't possible.)

The list in the batch file is all http links.

A file that triggered the first warning has a reference to a js file in the header. This uses 'script src="/_ScriptLibrary/***.js"'.

It appears that the linter is getting an incorrect mapping. It seems to be looking for a file path relative to the root of C, instead of an http path relative to the root of of the site.

I've checked the one file that I know tried to throw this. By eye the code looks good and CSE didn't identify any issues with the source, so I believe it is solid.

I don't know how many times this is thrown or by what pages (other than the one mentioned above).

Edit: Original content rendered badly by RSS engine. (Differed from online rendering.)

[Albert Wiersch]

Hello Mike,

I'm sorry for the delay. I believe I ran into this issue before. JavaScript Lint seems to try to follow js scripts and I don't think there is a way around this. CSE HTML Validator tries to ignore JavaScript Lint messages for other files... and I suspect JavaScript Lint could also calculate a path wrong in some circumstances and therefore could try to check a non-existent file.

If CSE HTML Validator is not filtering out a JavaScript Lint message saying that it could not open a file (like "can't open file: No such file or directory"), then it would be great if you could send me the source document to try to reproduce the problem so that CSE HTML Validator can filter that message out in the next update.

[MikeGale]

Hi Albert,

It was a bit of freak occurrence that I even spotted this.

When the jslint file couldn't be created/used the message leaked into the normal world of CSE validator.

When I tracked it down I didn't see anything in CSE itself. For thoroughness I checked the file name referred to in the previous error message. I was surprised that it had happened (silently) again, but it did let me track it down. I don't imagine other users would find it.

The visible fault is, unfortunately, not reproducible. I believe the "signature" of the causing page is simply an XHTML 1 transitional page with a script tag in the head, pointing to a root relative js file address.

The best I can suggest is checking the code that handles these jslint problems and seeing why it leaked through.

The file problem message was:

An error occurred trying to run the JavaScript linter. Error: 2005052001: can't open outputfile "C:\DOCUME~1\*UserName*\LOCALS~1\Temp\~cse_jsmsg43323888.tmp"

Hope that helps. If you need more let me know.

(It would be useful to modify the jslint code, but that, I'm sure, is too hard.)