Hello.
As you may have noticed, I've been working on HtmlHelp lately. Currently it should generally work, but still there is a lot to do. HTML should be correctly displayed and content tab should work. I was thinking about making more use of it in Wine. I think winecfg could benefit from it. We could easily add context help to winecfg using HtmlHelp API. Even just integrating a part of Wine User's Guide would be a good start. The main problem is building chm file (that is the file format of HtmlHelp). Microsoft has a HtmlHelp compiler (hhc) for it. We'd need something similar in Wine. AFAIK there is no good open source replacement. HtmlHelp maker (hhc, see [1]) doesn't create internal files (parts of chm that describes stuff like index or default topic) and it's GPLed so it's useless for Wine (unless author would relicense it for us). I think it would be a good project for Google Summer of Code. The task would be to write a hhc replacement and add a help option to winecfg. hhc replacement (say whhc) would have to be a plain UNIX tool (it means a bit of code duplication with itss.dll, just like we do in widl) so we could use it during compilation. I think its difficulty is good for SoC. Compressing code may be integrated from some other project. The remaining parts are code of parser of files describing chm and a little winecfg hacking. What do you think?
Jacek
* On Sun, 4 Mar 2007, Jacek Caban wrote:
HtmlHelp maker (hhc, see [1]) doesn't create internal files (parts of chm that describes stuff like index or default topic) and it's GPLed so it's useless for Wine (unless author would relicense it for us).
Quoting Paul Wise resume from the [1]:
| Started off hacking on frhed (a windows hex editor) | Made some lame/unfinished patches for POV-Ray | Wrote some tiny windows software | Wrote chmdeco/hhm/chmspec | Been hacking on activismo for a while | Looking forward to getting ADSL & contributing to xCHM, possibly WINE, GNOME, random projects. | I'd also like to become a debian developer at some stage.
Paul seems to be excited about Wine, so hhm relicensing should sound OK for him, IMHO. But only if proved it isn't breaking M$ licences [2], right?
I think it would be a good project for Google Summer of Code. The task would be to write a hhc replacement and add a help option to winecfg. hhc replacement (say whhc) would have to be a plain UNIX tool (it means a bit of code duplication with itss.dll, just like we do in widl) so we could use it during compilation. I think its difficulty is good for SoC. Compressing code may be integrated from some other project. The remaining parts are code of parser of files describing chm and a little winecfg hacking. What do you think?
Nice idea. Maybe even Paul himself would join that, as he states in his blog [3] he needs "Employment, bux, moolah, work". (I've Bcc-ed Paul in his message)
[2] http://msdn.microsoft.com/library/en-us/htmlhelp/html/hwhtmlhelpend-userlice... [3] http://bonedaddy.net/pabs3/
[please CC me on replies, I'm not subscribed to wine-devel]
On Mon, 2007-03-05 at 19:30 +0200, Saulius Krasuckas wrote:
- On Sun, 4 Mar 2007, Jacek Caban wrote:
HtmlHelp maker (hhc, see [1]) doesn't create internal files (parts of chm that describes stuff like index or default topic) and it's GPLed so it's useless for Wine (unless author would relicense it for us).
Quoting Paul Wise resume from the [1]:
Eeep, haven't updated that in years. Just deleted it and replaced with a link to my homepage.
Paul seems to be excited about Wine, so hhm relicensing should sound OK for him, IMHO.
I've previously offered to relicence the hhm code for inclusion in both chmlib and libmspack, but it seems neither project took up this offer. I don't think the hhm code belongs anywhere other than in one of them. The current version was more of a proof-of-concept, I wanted to push the code into a chm reading/writing library. IIRC, wine's itss.dll uses libmspack, so perhaps it should go there?
Also, if it does get included, I'd *STRONGLY* request/suggest that the code be modified to also include a README.hhm.txt (or similar) file saying something like "This file was not produced by Microsoft's HtmlHelp compiler, it may have incompatibilities preventing it from being used on Windows." I think this is an important inclusion in any non-MS compiler, one which I haven't got around to adding to hhm yet.
I've also previously (2003) pointed wine-devel at chmspec and offered to relicence chmdeco code for use in a wine version of hh.exe. No-one was interested though, saying that a CHM viewer was already implemented and then deteriorating into a thread about which browser to embed and forking KHTML and Mozilla and so on.
But only if proved it isn't breaking M$ licences [2], right?
The licence says this:
- Limitations on Reverse-Engineering, Decompilation, and Disassembly.
You may not reverse- engineer, decompile, or disassemble the SOFTWARE PRODUCT, except and only to the extent that such activity is expressly permitted by applicable law notwithstanding this limitation.
As written in the chmspec colophon, I didn't do any disassembly of Microsoft stuff, only black-box format discovery. Mainly this was by sending different inputs to Microsoft's hhc and observing the output, as well as looking at existing samples of CHM files. I imagine that is probably "reverse-engineering" according to Microsoft (although they haven't contacted me at all, so maybe not).
http://chmspec.nongnu.org/latest/Colophon.html
Nice idea. Maybe even Paul himself would join that, as he states in his blog [3] he needs "Employment, bux, moolah, work". (I've Bcc-ed Paul in his message)
I'm not a student, so I'm not eligible to implement that project as part of the SoC. If people want to pay me privately, that would be great since I'm in need of work at the moment. I'm happy to answer questions that come up in the SoC in relation to chmspec though.
Frankly I haven't done anywhere near enough reverse engineering of the internal CHM formats to even be close to thinking about implementing a full-blown CHM compiler that would create correct CHMs. There is enough info for CHM readers like chmdeco, xCHM and so on to be written, but not for compilers. Patches to chmspec are welcome though :)
Why does wine want a CHM compiler anyway? Can't Microsoft's one be used? Microsoft's one isn't distributed with the OS anyway.
Also, Matthew's ITSF (the container format built and read by itss.dll) spec has a few bugs that I did a bit of work on while I was sailing, haven't done anything recently on them though, I'm hoping to get to that soon.
Also, there is some bug in the chmlib/hhm/lzx_comp combination that causes xCHM and other chmlib based CHM viewers to crash when accessing CHMs created using hhm (linked against lzx_comp), this needs investigating and hhm/lzx_comp should not be used until it is.
--- Paul Wise pabs3@bonedaddy.net wrote:
The licence says this:
- Limitations on Reverse-Engineering,
Decompilation, and Disassembly.
You may not reverse- engineer, decompile, or
disassemble the SOFTWARE
PRODUCT, except and only to the extent that such
activity is expressly
permitted by applicable law notwithstanding this
limitation.
As written in the chmspec colophon, I didn't do any disassembly of Microsoft stuff, only black-box format discovery. Mainly this was by sending different inputs to Microsoft's hhc and observing the output, as well as looking at existing samples of CHM files. I imagine that is probably "reverse-engineering" according to Microsoft (although they haven't contacted me at all, so maybe not).
That sounds like the same way wine and samba are developed, which is called "clean-room reverse-engineering", and which should be legal. If it were not, using this code in wine wouldn't make anything worse ;-)
regards, Joris
____________________________________________________________________________________ Be a PS3 game guru. Get your game face on with the latest PS3 news and previews at Yahoo! Games. http://videogames.yahoo.com/platform?platform=120121
Hi Paul,
Paul Wise wrote:
Paul seems to be excited about Wine, so hhm relicensing should sound OK for him, IMHO.
I've previously offered to relicence the hhm code for inclusion in both chmlib and libmspack, but it seems neither project took up this offer. I don't think the hhm code belongs anywhere other than in one of them. The current version was more of a proof-of-concept, I wanted to push the code into a chm reading/writing library. IIRC, wine's itss.dll uses libmspack, so perhaps it should go there?
Yes, it should go to itss.dll to support writing mode in storages. But we can't use itss in hhc replacement as it has to be useful for Wine compilation. It means that we need plain UNIX app. That's the code duplication I wrote about.
Also, if it does get included, I'd *STRONGLY* request/suggest that the code be modified to also include a README.hhm.txt (or similar) file saying something like "This file was not produced by Microsoft's HtmlHelp compiler, it may have incompatibilities preventing it from being used on Windows." I think this is an important inclusion in any non-MS compiler, one which I haven't got around to adding to hhm yet.
Then we could add such README about whole Wine :-) Our aim is to improve things to be useful, not to write something and tell users to not use it because it sucks...
I've also previously (2003) pointed wine-devel at chmspec and offered to relicence chmdeco code for use in a wine version of hh.exe. No-one was interested though, saying that a CHM viewer was already implemented and then deteriorating into a thread about which browser to embed and forking KHTML and Mozilla and so on.
Things have changed. Now Wine has MSHTM/WebBrowser implementation that make it possible to support HtmlHelp.
But only if proved it isn't breaking M$ licences [2], right?
The licence says this:
- Limitations on Reverse-Engineering, Decompilation, and Disassembly.
You may not reverse- engineer, decompile, or disassemble the SOFTWARE PRODUCT, except and only to the extent that such activity is expressly permitted by applicable law notwithstanding this limitation.
As written in the chmspec colophon, I didn't do any disassembly of Microsoft stuff, only black-box format discovery. Mainly this was by sending different inputs to Microsoft's hhc and observing the output, as well as looking at existing samples of CHM files. I imagine that is probably "reverse-engineering" according to Microsoft (although they haven't contacted me at all, so maybe not).
That's fine for sure. Thanks for your work, it was very helpful to me!
Frankly I haven't done anywhere near enough reverse engineering of the internal CHM formats to even be close to thinking about implementing a full-blown CHM compiler that would create correct CHMs. There is enough info for CHM readers like chmdeco, xCHM and so on to be written, but not for compilers. Patches to chmspec are welcome though :)
I have to disagree. Surely it's enough to implement a compiler that will work with Wine's HtmlHelp. Also it should be possible to implement a compiler that will work with native hhctrl.ocx. Even if not everything is documented, dummy values should be good enough for most uncovered things. Also more things that will be needed may be discovered during SoC.
Why does wine want a CHM compiler anyway? Can't Microsoft's one be used? Microsoft's one isn't distributed with the OS anyway.
If we want to include chm files to Wine (like help for winecfg) we have to be able to compile them. Perhaps some apps ported with winelib would benefit from chm compiler.
Thanks, Jacek
On Tue, 2007-03-13 at 20:13 +0100, Jacek Caban wrote:
Yes, it should go to itss.dll to support writing mode in storages. But we can't use itss in hhc replacement as it has to be useful for Wine compilation. It means that we need plain UNIX app. That's the code duplication I wrote about.
AFAIK, itss.dll doesn't know anything about internal CHM formats, it just deals with the InfoTech Storage format (container format for CHMs), so no need for code duplication at all.
For a wine version of hha.dll/hhctrl.ocx and so on, there is always the possibility of creating a libhtmlhelp or something like that that can read and write the internal CHM formats, that uses a write-enabled libmspack or chmlib. Then use that library in wine's version of hha.dll/hhctrl.ocx/etc as well as the compiler, no need for code duplication at all. The current chmdeco code could form the initial decoding part of that library and write support could form around that.
Also, if it does get included, I'd *STRONGLY* request/suggest that the code be modified to also include a README.hhm.txt (or similar) file saying something like "This file was not produced by Microsoft's HtmlHelp compiler, it may have incompatibilities preventing it from being used on Windows." I think this is an important inclusion in any non-MS compiler, one which I haven't got around to adding to hhm yet.
Then we could add such README about whole Wine :-) Our aim is to improve things to be useful, not to write something and tell users to not use it because it sucks...
To be clear, this file would not be visible to users unless they unpack the files with extract_chmLib or istorage.exe or something. The idea was to document the fact that those files are non-Microsoft, so if there are issues with those files later, the person investigating those issues will immediately know that the file is non-Microsoft and then be able to direct any bug reports to the appropriate place. So it wouldn't be telling users not to use it, it would be telling recipients of hhm-created files where to go in case of problems.
Things have changed. Now Wine has MSHTM/WebBrowser implementation that make it possible to support HtmlHelp.
That is good to hear.
I have to disagree. Surely it's enough to implement a compiler that will work with Wine's HtmlHelp. Also it should be possible to implement a compiler that will work with native hhctrl.ocx. Even if not everything is documented, dummy values should be good enough for most uncovered things. Also more things that will be needed may be discovered during SoC.
I suppose it may be possible, I'd still suggest adding a README.hhm.txt to CHM files produced by hhm to warn that it may still be using dummy values. If you want to implement that I'm happy to provide explanations of any unclear bits in chmspec and accept chmspec bugs or provide other help on the project.
If we want to include chm files to Wine (like help for winecfg) we have to be able to compile them. Perhaps some apps ported with winelib would benefit from chm compiler.
Ok, that makes sense. I don't really understand the desire to use CHM files in Wine, that isn't relevant to this conversation though.
-
Jacek Caban -
Joris Huizer -
Paul Wise -
Saulius Krasuckas