Testing the new forum code (still as ugly, now with half-assed OOP!)

---

Testing the new forum code (still as ugly, now with half-assed OOP!)

Reply from old version.

EDIT

---

Testing the new forum code (still as ugly, now with half-assed OOP!)

Reply from old version.

EDIT

No regressions so far.

---

Testing new code.

And testing edits

---

Kathryn :o)

Well it's really the DB code that changed.

---

See http://boinc.berkeley.edu/trac/ticket/93

Thanks for the link to the ticket! Any way I can add my comments to the ticket?
TIA,
Ron

Should this be allowed? I replied to a post that doesn't even belong to this thread.

---

How the heck did you do that?

Editing the URL. Click reply, change the number after post= on the address bar to some random number.

---

Just a testing post for umlauts:

AE: � UE: � OE: �

The translation side of things is broken somehow. This is especially important, as now for example posts from Germany are made with a latin-1 charset:

When the translation is fixed again (UTF-8) the umlauts displayed in this message (written in latin-1) will be broken.

Here are two images, that this forum is also affected:

This quote manually inserted:

Editing the URL. Click reply, change the number after post= on the address bar to some random number.

Grrrrrrrrrrrr (database is being slow...)


I can't remember if we asked before (and the forums are too touchy to actually search) if you're using a password on resume.

And remind us what OS.

It works this way, is it a possibility to resurrect modded posts? *evil grin*

Edith says:
And some Umlauts as well:

���h, ich h�tte mal 'ne sch�ne Frage: W�rde das so reichen?

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

And some Umlauts as well:

���h, ich h�tte mal 'ne sch�ne Frage: W�rde das so reichen?

Nope, sadly not. You have also written the umlauts in latin-1. just look at your Signature and you will see what I mean. Your signature is encoded in UTF-8. The temporary fix is easy for the moment (I have also written to the mailing list). just use tr instead of tra in util.inc for the charset.

... just use tr instead of tra in util.inc for the charset.

Wouldn't it be better to translate it into HTML character entities before it goes into the database?

Then no one would ever have to care about which character set matches which page and language setting.

testing something

---

��u�

Testing the same with home.php switched to UTF8

---

��u�

Saenger, when you edit your board preferences, switch to ISO-8859-2 and then fix your sig :-)

It has never been a good idea to use different charsets for the NL support :-/

I already had a partial fix for that but trashed it because no one has been interested :-/

http://boinc.berkeley.edu/dev//forum_thread.php?id=990

(I put the source snippet back, it's based on outdated source of course but the source line it affects has not been changes since then)

... just use tr instead of tra in util.inc for the charset.

Wouldn't it be better to translate it into HTML character entities before it goes into the database?

Definitely not. The database must have the raw data, to be adapted to HTML or whatever on display. They did that (HTML entities on DB) with team names, and everybody started getting strange &crap; on BOINC Manager.

Plus, if charsets work correctly, only < > & " would need to have entities.

---

Saenger, when you edit your board preferences, switch to ISO-8859-2 and then fix your sig :-)

Been there, done that :)
THX!

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

Wouldn't it be better to translate it into HTML character entities before it goes into the database?

Definitely not. The database must have the raw data, to be adapted to HTML or whatever on display. They did that (HTML entities on DB) with team names, and everybody started getting strange &crap; on BOINC Manager.

Plus, if charsets work correctly, only < > & " would need to have entities.

You're right, HTML characters would look a bit crappy in the GUI :-/


The problem that I see with profiles and sigs is, that those depend on the charset of the creator whereas the charset of the page is given by the accepted language settings of a visitor or the profile language setting of a user and depends on the charset used by the translator (.po files).

So there will be either different conversions all the time or the stored format needs to work on all pages. Out of the formats that would match those requirements, PHP (without additional libraries) can handle HTML character attributes or unicode.


The code snippet I mentioned above deals only with the translations and should make it easier to use only one standard charset for all PHP pages, independant from the charset that has been used for the .po files.

This would allow all users to have unicode on their profile pages, so those would match the page where UOTD is printed as well as the forum threads.


The modification will not be enough to solve all problems at once but it should make it much easier to deal with the problems that are left.

The problem that I see with profiles and sigs is, that those depend on the charset of the creator whereas the charset of the page is given by the accepted language settings of a visitor or the profile language setting of a user and depends on the charset used by the translator (.po files).

The code snippet I mentioned above deals only with the translations and should make it easier to use only one standard charset for all PHP pages, independant from the charset that has been used for the .po files.

All translation files have already been modified to use exclusively UTF-8. I don't know if more recent upgrades brought it back to a different charset...

---

Oh, you're right. This page is ISO-8859-2 for me in SeaMonkey (beeing logged in) and Windows-1252 in IE (not beeing logged in) - that's why I assumed that there were still multiple charsets in the .po files.

p.s.: Saenger's sig looks correct in both btw., but why has his profile edit page been UTF-8 before ?

Oh, you're right. This page is ISO-8859-2 for me in SeaMonkey (beeing logged in) and Windows-1252 in IE (not beeing logged in) - that's why I assumed that there were still multiple charsets in the .po files.

p.s.: Saenger's sig looks correct in both btw., but why has his profile edit page been UTF-8 before ?

The charset is defined through the language files. For German and English this is UTF-8. As it is not specified anymore (because tra("CHARSET"); doesn't work anymore) your browsers standard charset is used. So by fixing your signature, which was written in UTF-8 correct, you will break it, when the translation function works correct again.

Bye,
Timo!