AntConc 3.4.0 for (Native) Windows, OS X, and Linux

[Laurence Anthony]

Dear All,

I have now released AntConc 3.4.0 for Windows, OS X, and Linux. This is a semi-major update that incorporates a few changes that I was making before starting development on the Python-based AntConc 4.0. You can find the new version at the following URLs:

http://www.antlab.sci.waseda.ac.jp/software.html
http://www.antlab.sci.waseda.ac.jp/antconc_index.html

From this release, version numbering (x.y.z) will reflect the following: x => major release, y => minor improvements + bug fixes, z => bug fixes only. This version has been quite heavily tested in preparation for its inclusion in the Lancaster University MOOC online corpus linguistics course (https://www.futurelearn.com/courses/corpus-linguistics). However, no doubt, a few minor issues remain. If you notice anything, please send me a note and I can remedy the problem very quickly. You may not notice many differences in 3.4.0, but a few highlights are below (taken from the Readme page).

I hope you find the new version to be an improvement and that it is useful in your research and studies.

Best regards,
Laurence.

Highlights for AntConc 3.4.0

New Features:
1.    The Concordancer tool output now correctly centers the search terms regardless of language. This is especially useful for Asian languages such as Japanese, Chinese, and Korean that have two-byte character widths. Note that this is achieved by using two columns in the 'KWIC' pane.  If you click CTRL-A in the pane, only one of these hidden columns will be selected. Use ALT-A to select all columns.

2.    Line breaks in the Concordance Tool output can now be replaced by a user defined character (default being a single space). Again, this is especially useful for Asian languages because traditional replacement by a single space might inappropriately split words that cross line-break boundaries.

3.    The Collocates Tool no longer includes the center word in the collocation table. This results in more meaningful collocate values.

4.    Not exactly a feature, but Concordance Tool searches are now only centered on completion rather than after finding the first line. This is generally a more pleasant experience.

5.    Whitespace options in the wildcards menu have now been extended.

6.    A database now runs in the background and is used to store KWIC and Plot concordance results. This alleviates out of memory problems with earlier versions and also allows much faster sorting of the KWIC. The new database model also improves the performance of general KWIC and Plot searches.

7.    In the Advanced Search, if you add multiple context words, AntConc will look for *any* of these in surrounding context of the search term rather than *all* of them (as in earlier versions). This change has been implemented as a result of many requests for this function. I hope it does not inconvenience too many people.

8.    In the Collocates tool, if you click on a word, if will search for that word in the Concordance Tool but restrict results to only those where the original search term is in the neighborhood of the collocate as set by the Collocates tool. Again, this change has been implemented as a result of many requests for this function.

9.    Colors for the highlight and ordered words in the KWIC display have been changed to be a little easier to see.

10.    The font size for the main results is now a little larger.

Bug fixes:
1.    The Collocates Tool sometimes use to report values of "-1" (when the only collocates of the center word was itself) and "-2" (when the collocate value was not able to be calculated due to an out-of-sync word list). The first case ("-1") no longer exists (see feature 3 above). In the second case ("-2"), the value has now been replaced with "0" to avoid confusion.

2.    In the Word List Tool, Alt-A now correctly selects all the panes including the Lemma list.

3.    When no corpus file names were selected in the left panel, sometimes the “Close file” menu option would delete the first file name in the list instead of reporting that no file was selected. The has been fixed.

4.    After changing the size of the Search Entry font option in the global settings, the new font size would not be reflected in the sample display. This is now fixed.

5.    If the "invert sort" option was activated in any tool with this feature, an "Array XXX" error was produced in the results window. I apologize for this very inconvenient bug.

6.    Several other minor bugs including one that sometimes led to tags not being hidden properly after applying tag settings have been fixed.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

Dear All,

I just heard from a user that had problems starting the new AntConc 3.4.0 release with their old settings file in the same folder. This will not work. Please remember to delete your old settings file before starting the program. You can then export a new version of the settings file from the file menu.

A note about this has been added to the download pages.

Best regards,

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Gian Paolo Renello]

Hi Laurent,
thank you for your continuous work on this wonderful program. My students use it woith great interest.

For what concern this new version i have to remember all 64-bit system users that AntConc usually will not work on a 64 bit system unless they intall a lib32 file. It is a 32-bit FreeType-based font drawing library for X. For archlinux users it is enough to install lib32-libxft (sudo pacman -S lib32-libxft) and that's all. I guess for Ubuntu and derivates will suffice to install libxft2 or something like that (32 bit version of course), or, eventually, look for a freetype drawing library for X in a search engine

Regards
Gian Paolo Renello

[Markus Killer]

Hi Gian Paolo,

I can confirm that the minimal requirements on a 64-bit *buntu variant are the following three 32-bit libraries:

sudo apt-get install libx11-6:i386 libxss1:i386 libxft2:i386

Best wishes,

Markus

[Mura Nava]

this is great thanks

there is still the screen refresh issue in Mavericks OSX i.e. when you say change a parameter the screen does not show change though it has been changed if say you switch tabs and come back to original tab

[Laurence Anthony]

Yes, I know about this. I'm still waiting for ActiveState to update their version of Perl. Once they update it, AntConc 3.4 will run fine in Mavericks. You'll see I have a note on the download page. At the moment, Mavericks users will have to go back to the classic AntConc version 3.2.4.

Laurence.

--
You received this message because you are subscribed to the Google Groups "AntConc-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to antconc+u...@googlegroups.com.
To post to this group, send email to ant...@googlegroups.com.
Visit this group at http://groups.google.com/group/antconc.
For more options, visit https://groups.google.com/groups/opt_out.

[Reed Darsey]

On 25 Jan 2014 at 12:15, Laurence Anthony wrote:

> I have now released AntConc 3.4.0 for Windows, OS X, and Linux.

Thank you for this new release.

For the Windows version, please be aware that some users have their
display settings at higher than 100% to make the screen easier to
read. E.g., I use 125%, which cuts of the end of the text in the
tabs, and the boxes for Global Settings - User-Defined Token Class.

[Umut Demirhan]

Dear Laurence,

 

Thank you for the new release. It looks many things have been changed in the backend process.

 

I would like to know one thing. As far as I understand, the new concordance tool generates the concordances within +-5 window span by default; however, the “Search Window Size” option still shows the span as “characters”. Do you confirm that from now on, we will use words as a span entry?

If yes, there is a problem with “Search window size” because it still shows the “character number” as an entry,  If not, what should we use for the span entry? +-5 words or +-50 characters

 

Best regards,

Umut

 

--

[Reed Darsey]

On 25 Jan 2014 at 12:52, I wrote:

> For the Windows version, please be aware that some users have their
> display settings at higher than 100% to make the screen easier to
> read. E.g., I use 125%, which cuts of the end of the text in the
> tabs, and the boxes for Global Settings - User-Defined Token Class.

And I can't see the Max. N-Gram Size.

[Reed Darsey]

On 25 Jan 2014 at 12:15, Laurence Anthony wrote:

> However, no doubt, a few minor issues remain. If you notice anything,
> please send me a note and I can remedy the problem very quickly

File View stopped working after a few times. (I'm running Windows 7
Home Premium 64-bit, and twiddling with the options quite a bit.)

[Laurence Anthony]

Dear Reed Darsey,

AntConc 3.4.0 works fine with magnified display settings (e.g. 125%). Just resize the AntConc window. This is the same as with the older 3.3.5.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

Dear Reed Darsey,

As I wrote earlier, AntConc 3.4.0 works fine with magnified display settings (e.g. 125%). Just resize the AntConc window. This is the same as with the older 3.3.5. The default setting has a very small windows size of just 800x600, which is needed for older monitors often found in classrooms. Normally, you will be using a much higher monitor resolution, so you can resize AntConc to a much bigger size. At 125% the User-Defined Token Class settings box is a little tight, but it should still show properly.

Laurence.

 

[Laurence Anthony]

Hi Umut,

 

Thank you for the new release. It looks many things have been changed in the backend process.

I hope that's a statement about the *positive* changes with the new release!
 

 

I would like to know one thing. As far as I understand, the new concordance tool generates the concordances within +-5 window span by default; however, the “Search Window Size” option still shows the span as “characters”. Do you confirm that from now on, we will use words as a span entry?

If yes, there is a problem with “Search window size” because it still shows the “character number” as an entry,  If not, what should we use for the span entry? +-5 words or +-50 characters

In the KWIC concordance tool, spans are still determined based on the *number of characters* specified in the "Search Window Size" option. However, because AntConc now stores the results in a database, it takes that span, tokenizes the results based on the token definition, estimates how many database columns are needed to store those tokens, and then saves the tokens in the database, trimming certain lines down to the correct number of columns where necessary. So, if you increase the "Search Window Size" to say 100 (characters), you'll see that the context increases. The important point to note is that the initial search is still based on a *number of characters* span.

I hope that helps.

Laurence.

 

[Laurence Anthony]

On Sun, Jan 26, 2014 at 5:13 AM, Reed Darsey <Re...@mchsi.com> wrote:

Dear Reed Darsey,

As I wrote earlier, AntConc 3.4.0 works fine with magnified display settings (e.g. 125%). Just resize the AntConc window. This is the same as with the older 3.3.5. The default setting has a very small windows size of just 800x600, which is needed for older monitors often found in classrooms. At 125% the User-Defined Token Class settings box is a little tight, but it should still show properly.

Laurence.

 

[Reed Darsey]

On 26 Jan 2014 at 8:15, Laurence Anthony wrote:

> AntConc 3.4.0 works fine with magnified display settings (e.g. 125%).
> Just resize the AntConc window. This is the same as with the older
> 3.3.5.

What threw me was that the older AntConc remembered the window size
between sessions, and 3.4.0w doesn't. I must've set the older one a
long time ago, and forgot about it, then tried to adjust the Token
Definition window, which doesn't resize.

Another problem that likely only affects me is that I tried
N-gramming some very large files that I had done in 3.3.5, but they
crashed 3.4.0. I guess as progress is made toward 4.0 and 64-bit,
this won't be an issue.

[Laurence Anthony]

Hi Reed,

I have not changed a single line of code related to Ngrams. So, I'm not sure why it would crash now.

Can you confirm that the same Ngram search in 3.3.5 now crashes in 3.4.0. This is quite odd.

Laurence

[Laurence Anthony]

Hi,

Thanks for the feedback.

When you say "it stopped working", can you explain what you mean? Did the software freeze, or just show no results, or something else.

Also, does it do this always after a fixed number of correct searches?

The clearer your explanation is, the quicker I can probably solve it.

Laurence

[Reed Darsey]

On 26 Jan 2014 at 8:42, Laurence Anthony wrote:

> I have not changed a single line of code related to Ngrams. So, I'm
> not sure why it would crash now.

A change you didn't state in the change log is that the N-Grams now
disregards line breaks, so that the example I keep using of:

The cat walks the dog. The
bird flies from its cage.

now gets "the bird".

So, its likely that the large files that 3.3.5 could do overwhelm the
underlying engine in 3.4.0.

The "same" N-gram operation in 3.4.0 won't be the same as in 3.3.5.

If I want top get the same effect, I have to:

(1) Tool Preferences - Concordance set Line break replacement to "%".
(2) Global -Token - Custom put the "%" in User-Defined.

The crash is what Windows reports, that the application has stopped
working. One time before that, a error box said something about a
Visual C or something error, and to alert the developer.

[Reed Darsey]

On 25 Jan 2014 at 18:30, I wrote:

> A change you didn't state in the change log is that the N-Grams now
> disregards line breaks, so that the example I keep using of:
>
> The cat walks the dog. The
> bird flies from its cage.
>
> now gets "the bird".

More detail:

AntConc 3.3.5:

#Total No. of N-Gram Types: 9
#Total No. of N-Gram Tokens: 9
1 1 1 bird flies
2 1 1 cat walks
3 1 1 dog the
4 1 1 flies from
5 1 1 from its
6 1 1 its cage
7 1 1 the cat
8 1 1 the dog
9 1 1 walks the

AntConc 3.4.0:

#Total No. of N-Gram Types: 10
#Total No. of N-Gram Tokens: 10
1 1 1 bird flies
2 1 1 cat walks
3 1 1 dog the
4 1 1 flies from
5 1 1 from its
6 1 1 its cage
7 1 1 the bird
8 1 1 the cat
9 1 1 the dog
10 1 1 walks the

[Laurence Anthony]

Hi Reed,

Ah yes. I now see why you are getting differences. I can also see why the N-Gram tool might now crash with very large files.

In the old version (3.3.5), all the tools worked in a slightly different way regarding line breaks. The KWIC tool would automatically replace line breaks with spaces when producing results (this is fairly standard in the field). However, when a KWIC result was clicking on, AntConc would jump to the File View tool, but results crossing line breaks would not be found. So, you would have the strange case of a result appearing in the KWIC tool and not appearing in the File View tool.

Similarly, the Clusters/N-Grams tool would only search for clusters/n-grams per line. As a result, the number of hits for a pattern in the Clusters/N-Grams tool would not match the number of hits for the same pattern in the KWIC tool. Many people asked why clicking on a result in the Clusters/N-Grams tool did not produce the same number of hits in the KWIC tool.

To rectify this problem, I added a new option to specify exactly what a whitespace character meant in different tools (the default is now "one or more non-tokens"). This makes the KWIC and File View tool results consistent. I also changed the way that files are read into AntConc so that line breaks are treated in the same way. Now, both the KWIC tool and Clusters/N-grams tool read in a complete file, replace line breaks (according to the line break replacement option), and then process the files. This results in matching hits across both tools.

I think the new approach is much more consistent, but it does introduce the problems that you have mentioned. I think the following revisions might be good:

1) Add a line break option to the clusters/N-grams tool preferences so that users can decide whether or not the tool looks for clusters/N-grams across line breaks or not. This introduces the danger of the setting becoming out of sync with the one for the KWIC tool. But, at least it would highlight why the results might not match in certain cases.

2) Edit the N-Gram processing script so that if line breaks are *not* ignored, the processing will go line by line saving memory on very big files. This will allow your N-Gram searches to work again.

Does that make sense?

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

On Sun, Jan 26, 2014 at 8:30 AM, Reed Darsey <Re...@mchsi.com> wrote:

The crash is what Windows reports, that the application has stopped
working. One time before that, a error box said something about a
Visual C or something error, and to alert the developer.

Hi again,

I suspect this crash might be due to running out of memory when trying to process very big files. Such crashes should no long occur in the KWIC tool (due to the introduction of a database), but they might still happen in the other tools that still rely on slurping in an entire file before processing. Can you remember what you were doing when the crash happened?

Laurence.

[Reed Darsey]

On 26 Jan 2014 at 10:31, Laurence Anthony wrote:

> 1) Add a line break option to the clusters/N-grams tool preferences
> so that users can decide whether or not the tool looks for
> clusters/N-grams across line breaks or not.

This was a suggestion I've made during these discussions in the past.

Some text I have is already one sentence per line break, so AntConc
works like I want.

For other text, my plan was to explore "sentence boundary
disambiguation" methods to pre-process text into one sentence per
line break.

I'm aware that others might want the line breaks to be ignored, to
match other software's behavior.

But, I hadn't planned on the behavior to change without the choice.

> 2) Edit the N-Gram processing script so that if line breaks are *not*
> ignored, the processing will go line by line saving memory on very big
> files. This will allow your N-Gram searches to work again.

Program should alert the user that this is happening, though.

> Does that make sense?

Well, the much older kfNgram handles these 20 - 40 MB files in less
than a minute, so I thought that as AntConc develops, big file
handling wouldn't be an issue anymore.

[Laurence Anthony]

Hi Reed,

I'm adding the functionality right now. It will appear as a development release later today. (Are you on Windows?). I see how the new release can be an inconvenience. Sorry to forget to add the option in the 3.4.0 release!

Although kfNgram is an older program, it does N-Gram processing in a very smart way. I've actually spoken to the developer with the idea of incorporating its algorithm into AntConc. kfNgram creates lots of temporary N-gram text files and then merges the results. It should be possible for AntConc to do the same. At the moment, AntConc simply does everything in memory, which is fast, but not at all appropriate for big files.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Reed Darsey]

On 26 Jan 2014 at 11:00, Laurence Anthony wrote:

> I'm adding the functionality right now. It will appear as a
> development release later today. (Are you on Windows?). I see how the
> new release can be an inconvenience. Sorry to forget to add the
> option in the 3.4.0 release!

The options should be explained in the change log, so users are made
aware of the differences between the versions, to not get counfused
by getting different results all of a sudden.

> Although kfNgram is an older program, it does N-Gram processing in a
> very smart way. I've actually spoken to the developer with the idea
> of incorporating its algorithm into AntConc. kfNgram creates lots of
> temporary N-gram text files and then merges the results. It should be
> possible for AntConc to do the same. At the moment, AntConc simply
> does everything in memory, which is fast, but not at all appropriate
> for big files.

Using temporary files might be the safest, but on a 16 GB 64-bit
system, there should be plenty of memory to work on any reasonable
file in memory. (I know I'm working with unreasonably big files, but
want to let you know, in case future users need to also work on big
files.)

[Laurence Anthony]

Dear Reed,

> I'm adding the functionality right now. It will appear as a
> development release later today. (Are you on Windows?). I see how the
> new release can be an inconvenience. Sorry to forget to add the
> option in the 3.4.0 release!

The options should be explained in the change log, so users are made
aware of the differences between the versions, to not get counfused
by getting different results all of a sudden.

Good point. To be honest, I forgot about the impact of the changes on the Clusters/N-grams tool.  I can certainly add a note.

> Although kfNgram is an older program, it does N-Gram processing in a
> very smart way. I've actually spoken to the developer with the idea
> of incorporating its algorithm into AntConc. kfNgram creates lots of
> temporary N-gram text files and then merges the results. It should be
> possible for AntConc to do the same. At the moment, AntConc simply
> does everything in memory, which is fast, but not at all appropriate
> for big files.

Using temporary files might be the safest, but on a 16 GB 64-bit
system, there should be plenty of memory to work on any reasonable
file in memory. (I know I'm working with unreasonably big files, but
want to let you know, in case future users need to also work on big
files.)

Memory management of Perl programs on Windows in notoriously bad. As you say, with a 64-bit system and a lot of memory, it should still be possible to work with quite large files. Many people have asked for a 64-bit version of AntConc. Maybe now is the time to produce it.

Sorry for the inconvenience.

Laurence.

[Reed Darsey]

On 26 Jan 2014 at 11:57, Laurence Anthony wrote:

> Many people have asked for a 64-bit version of AntConc. Maybe now is
> the time to produce it.
>
> Sorry for the inconvenience.

AntConc is the most configurable and useful N-grammer out there, and
I'm pushing it beyond what it's reasonably meant to do.

[Laurence Anthony]

Not at all. I would love to have the N-gram tool work as well as kfNgram. Actually, I even started coding a completely separate N-gram tool to try things out. Let me work on this a little more. I would certainly be interested to see how well AntConc's N-gram tool works on a 64-bit system.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

Dear Reed Darsey and All,

Based on the earlier discussion regarding the new N-Gram tool implementation in AntConc 3.4.0, I've made a couple of changes and released a development version (3.4.1 dev) that adds a new menu option for the Cluster/N-Grams tool to allow you to either replace line breaks with a white space (or another character) or leave them as is. I think this is an elegant solution. See what you think.

If you notice other issues or problems, I'll address them by updating this development release and then release 3.4.1 when the timing is good.

You can find the latest development release here (currently only for Windows):
http://www.antlab.sci.waseda.ac.jp/antconc_index.html

Regards,
Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Reed Darsey]

On 27 Jan 2014 at 2:43, Laurence Anthony wrote:

> If you notice other issues or problems, I'll address them by updating
> this development release and then release 3.4.1 when the timing is
> good.

In 3.4.1 dev, the main screen doesn't remember the changed size, so
has to be re-sized every time I start the program.

In Global Settings - Fonts, my choice of Consolas for the Results
Font gets changed to Segoe UI. (The change is seen when the program
is exited and re-started.)

In 3.3.5w the above worked corrrectly.

[Reed Darsey]

On 26 Jan 2014 at 12:34, I wrote:

> In Global Settings - Fonts, my choice of Consolas for the Results
> Font gets changed to Segoe UI. (The change is seen when the program
> is exited and re-started.)

I forgot I need to use File - Export settings to make the changes
"stick."

But the main screen still needs to be resized each time.

[Reed Darsey]

On 27 Jan 2014 at 2:43, Laurence Anthony wrote:

> (3.4.1 dev) that adds a new menu option for the Cluster/N-Grams tool
> to allow you to either replace line breaks with a white space (or
> another character) or leave them as is. I think this is an elegant
> solution.

Yes, I now can get the same N-Gram results as I did in previous
versions.

[Mura Nava]

hi am following your vid on #corpusmooc and the advanced search window box does not appear in mavericks osx, prev version of antconc 3.3.5m window appears fine

On Saturday, 25 January 2014 17:11:35 UTC+1, Laurence Anthony wrote:

Yes, I know about this. I'm still waiting for ActiveState to update their version of Perl. Once they update it, AntConc 3.4 will run fine in Mavericks. You'll see I have a note on the download page. At the moment, Mavericks users will have to go back to the classic AntConc version 3.2.4.

Laurence.

On Sat, Jan 25, 2014 at 11:48 PM, Mura Nava <mura...@gmail.com> wrote:

this is great thanks

there is still the screen refresh issue in Mavericks OSX i.e. when you say change a parameter the screen does not show change though it has been changed if say you switch tabs and come back to original tab

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan

E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

--

[Laurence Anthony]

AntConc 3.3.5 and 3.4.0 are likely to generate odd problems in Mavericks. The problem is a refresh problem. If you move out of the window click a few times and move back in, things may update properly. But, the effect is very unpredictable.

The safer way to proceed on Mavericks is to use the classic version (3.2.4) until the problem is fixed by ActiveState, who create the Perl version that I use to make the software.

I'm really sorry about this, but there really isn't anything I can do (except move to Python!)

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

Hi Reed and All,

I've updated the development release of AntConc so that a settings file will now also remember the size and position of the main window.  Check out the readme file in the development version for a full list of changes.

You can download it from here:
http://www.antlab.sci.waseda.ac.jp/antconc_index.html

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Reed Darsey]

On 28 Jan 2014 at 22:02, Laurence Anthony wrote:

> I've updated the development release of AntConc so that a settings
> file will now also remember the size and position of the main
> window.  Check out the readme file in the development version for a
> full list of changes.

In build 3411, the main window does remember its size.

But, while the Global Settings - Token Definition window is now
re-sizable, so I fully see the User Defined Token Class boxes, after
dragging the lower right corner, when the screen in gone back into,
those lower boxes are not at all visible. The new screen size does
not stick.

[Laurence Anthony]

Hi Reed,

>after dragging the lower right corner, when the screen in gone back into, those lower boxes are not at all visible

I don't understand this point. Can you explain it again?

Note that I didn't intend making the settings windows stick in the the user settings. They are only opened occasionally so I though resizing each time would be fine.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Reed Darsey]

On 29 Jan 2014 at 9:27, Laurence Anthony wrote:

> >after dragging the lower right corner, when the screen in gone
>> back into, those lower boxes are not at all visible
>
> I don't understand this point. Can you explain it again?
>
> Note that I didn't intend making the settings windows stick in the
> the user settings. They are only opened occasionally so I though
> resizing each time would be fine.

*Nothing* shows under the text "User-Defined Token Class" -- just a
amall rectangle. A new user would have no idea that there is
anything there to set.

In other words, before, the lower half of the second line of "Use
Following Definition" was cut off. Now the "Use Following
Definition" and "Append Following Definition" text and boxes are not
even visible, unless the screen is made larger.

I know that those settings are there, and can enlarge the screen to
see them, but a new user might not know to enlarge that settings
screen.

[Laurence Anthony]

Hi Reed,

That clarifies things. When I use the magnified view here, I can see the problem. The "magnified screen font" feature of Windows is so annoying. It really messes up interface design. To allow everything to appear correctly with the magnified view and still restrict the minimum window size to 800x600 pixels, I'm going to have to redesign the settings page.

Ahh....

Laurence.

[Laurence Anthony]

Hi again Reed,

I updated the development release with a slightly modified settings window that hopefully address the problem you mentioned. I also heard of a few other issues on Macintosh that I fixed. Check out the new readme file. Everything is documented.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Reed Darsey]

On 29 Jan 2014 at 12:55, Laurence Anthony wrote:

> Check out the new readme file. Everything is documented.

Now I can see everything. I believe that many Windows users use the
125% screen magnification.

[Laurence Anthony]

Hi Reed,

You might be right. I thought I should try to address the issue immediately, especially as I'm coding a lot at the moment and everything is set up to make quick compiles for all three platforms.

Please keep the feedback coming. It's great.

Laurence.

###############################################################
Laurence ANTHONY, Ph.D.
Professor
Center for English Language Education in Science and Engineering (CELESE)
Faculty of Science and Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan
E-mail: antho...@gmail.com
WWW: http://www.antlab.sci.waseda.ac.jp/
###############################################################

[Laurence Anthony]

Hi Reed and All,

You'll notice 3.4.1 is now released. The new version addresses all the issues discussed in this thread. I hope you find it useful.

Regards,
Laurence.