On Wed, 23 Dec 2020 11:03:57 -0500 (EST) Dariusz Piatkowski wrote:
>
>Hi Gregg, everyone...
>
>On Sun, 13 Dec 2020 13:20:26 -0700 (MST) Gregg Young wrote:
>>Hi
>>
>>Attached is lSwitcher 2.93 RC 4. It hopefully fixes all of the font,
>>menu and layout issues
>>identified by Alfred.
>
>
>Attached please find the capture of the Taskbar and Global TABs from the
>Settings screen. These are done in both Polish (native speaker here, so
>there are a few suggestions I'd like to make about the translations, not
>sure how to go about it though, is that for the Translators list?) and
>English.
I received the Polish translation from Tae Wong and since I know nothing about Polish accepted it at face value. I am fine with you fine tuning it.
It isn't a ArcaOS project it is on netlabs. http://trac.netlabs.org/lswitcher The instructions for getting the SVN is on the above page. I can send the files to you if you prefer. I just committed all the language related stuff I have. There is no Polish help file.
>
>ENGLISH - looks good, the font issue I had previously reported is now
>gone.
>
>POLISH - the font switches to something else, not sure what it is, but
>certainly it is readable. I did not try switching the font to something
>else. My regular language config is to use the following setup in
>CONFIG.SYS : CODEPAGE=850,852.
OK the way this was done is the General (Misc) page is now Unicode so the languages in the drop down all have all the correct glyphs. The other pages are set to code page 852 internally. However when I tried it with Warp Sans it didn't have all the glyphs so I changed it to Times New Roman MT 30 which is the only font that ships with OS/2 that rendered everything. Perhaps there are other choices that will work for Polish I didn't try many others. It was clearly the only one that worked for Russian.
The object was so you could use any of the languages on any SB code page and have everything render correctly. If this doesn't happen what is the point of having the language selectable?
>
>Here the 852 codepage is meant to provide a secondary codepage support
>for the Polish language. Not sure if this has any bearing on what I'm
>seeing, but I figured it might be worth mentioning.
>
>
>>Still to do/decide on.
>>1. ùAdding undo to the settings?
>>A. make the hide button work as described in the help (ie only save
>>non-settings stuff to the ini
>>on close?
>>B. Implement a full undo? This is from an earlier email:
>>a. What do we intend for "undo" to actually do? Should it undo all the
>>changes on every page or
>>just the current page? Should it close the dialog or leave it open for
>>additional changes?b. How do we handle exclude lists do I undo the last
>>entry added/removed or do I need to
>>determine all the entries made in this session for removal or adding
>>back perhaps some of each
>>kind.
>>c. Significant changes of this kind add complexity and as such more
>>opportunity for defects. Is
>>having what will probably be a little used option worth this?
>
>
>I think you hit the nail right on the head with the last point, my
>thinking: too much complexity for a feature that'll be rarely used. Not
>saying it's unimportant, but I'd wager a guess that whoever needs to
>use this, especially if it's a one-off type of a situation (as opposed
>to constant use) really won't be getting the time savings required to
>justify the coding effort & time required to implement.
>
>Thanks once again Gregg for the awesome work!
Thanks
Gregg
PS: I saw your post re fat 32 and your test results don't make any sense. I am not disputing that the Netlabs version may be faster.
The issue is setting cache size to 0 in the netlabs version has no effect on the code. The cache there is disabled in the code. Also even if it was enabled one of the things I fixed was to make setting cache size to zero actually disable the cache. That wouldn't happen with the netlabs code.
Also based on testing using DFSee USB 2 tops out a ~3 mg/sec so the spikes don't make any sense either. (This test bypasses the ifs)
Just a note. The code I started with leaked memory and was probably not multi-processor safe.
back to list