Things noticed on first test of V 9 (8.9911)

[MikeGale]

I ran some batch tests on the new version.

1) When I checked for a new beta from the editor it reported none available but manual entry of strings downloaded it for me.

2) I ran two batch tests on previously validated content which sits on a staging server, IIS.

Batch test 1.

1) 91 targets, mostly aspx, a fair number of css and few htm's.

2) I got 22 pages with errors from these.

3) The first bunch were aspx's which had 1 or 2 additional characters appended to the real content of the file.

4) A css errored. From the report it had approximately doubled in size. All errors after the end of the real file. The messages suggested that a mangled version of the original file (maybe first few characters on each line missing) had been appended to the real file.

5) Another css errored. It was 36 lines long but reported errors between lines 165 and 211. Interestingly that range is similar to the size of the real file.

6) One aspx of 89 lines showed errors in lines 89 to 97. The details indicated that unique content from a file 9 positions earlier in the batch list was present. That earlier page had not been flagged with any errors.

7) I saw 91 docs, 22 errors and thought, that looks like 1 in 4. I didn't find a regularity in the order from the "check list" however. Errors were found at 1, 6, 12, 13, 25, 26, 32, 58, 59, 61, 66, 67, 68, 70, 72, 74, 75... which isn't an obviously regular pattern.

Link checks are good.

9) I noticed the progress indicators for pages and link checking. Really useful to see what's going on. (I haven't noticed those before, are they some of the changes you mention in the change log?)

10) These checks were on a freshly installed version, and happened immediately after it was started.

Batch test 2.

1) 9 files, all html.

2) 4 files came up with errors. All had a single additional character appended at the end of the file.

Maybe the checker is looking at text that sometimes exceeds the size of the real file. The latter part of that text, in one case, clearly came from another page in the check list. In a css file, the checked text may have been mangled a bit then appended to itself.

I have not run exhaustive checks to see where it all came from.

[Albert Wiersch]

I ran some batch tests on the new version.

1) When I checked for a new beta from the editor it reported none available but manual entry of strings downloaded it for me.

I think this was because I didn't immediately update the scripts that tell you a new version is available. They should be updated now.

Thanks for the reports on the extra text. I have some suspected ideas and places to look and will try to find the problem. I assume these issues are occurring in the Batch Wizard with URLs and not local files.

If you are able to provide a way for me to reproduce these problems on my end, then please let me know as that would be very helpful.

9) I noticed the progress indicators for pages and link checking. Really useful to see what's going on. (I haven't noticed those before, are they some of the changes you mention in the change log?)

The link checking progress is improved. Yes, it is mentioned in the history/change log. Glad you like it.

[MikeGale]

Ahead of that next build I ran some additional tests as follows:

1) Using last release of version 8, fairly long list. Did it twice. No issues.

2) Using 8.9911. Same as above. First time threw up errors. Second time bombed.

3) As for 2, results same.

In both cases a bug file with zip attached was generated. Both went into Outlook, but were not sent. I manually sent each.

This list, is on a staging server, but the same files are on a live server, so it's amenable to public testing, if needed.

I'll test with the special build now.

The exception reported was StrStoreW::SetStringFromUtf8():Utf8ToUnicode() failed. It froze the program and I needed Task Manager to kill the process both times.

[MikeGale]

Hi,

I've just run the special build. Only 2 runs, same batch, one after the other, identical to previous tests of 8.9911.

I'm glad to report that 8.9912 seems to be a great improvement. No problems found in first test.

1) It detected no errors (this is an errors only test). That's what I expect.

2) Both report sets (I've archived) are the same except for timing, time and in page anchor values.

Thanks.

Now I must get off to do something else. I'll will run more tests.

[MikeGale]

I have now run a series of additional batch tests.

1) Repeated tests in the same batch tool session.

2) Editing the batch file list.

3) Following links.

4) All error only thus far.

5) Html pages and css's.

6) A couple of different web sites.

7) Local content and remote content.

A few files in the list up to 90.

I have seen no further cases of "ghost content" added to pages, using 8.9912.

[Albert Wiersch]

I have seen no further cases of "ghost content" added to pages, using 8.9912.

Thanks Mike! Looks like I'll release a new BETA today.