-
In my Abbrevia4 redesign every archive provides an TArchStream for reading and writing data. The application code does not need to know that it reads or writes an archive, it can work with the archive, or with files in an multi-file archive, like with every other file.
2009-11-09 03:32:42 UTC by vbdis
-
Nicole,
That does clear things up. Using ExtractToStream/AddFromStream with a stock Abbrevia definitely won't be the best way to handle things.
The first problem is FDICopy, which ExtractToStream uses to actually get the file contents. FDICopy doesn't maintain state from call to call, so it has to re-read the CAB header and seek into the archive for each file you're extracting. If you...
2009-11-09 03:15:06 UTC by capeterson
-
Also, I agree that the lack of a TDeflateStream/TInflateStream like zlib.pas provides is an issue, but looking at the code, I don't think it will be an easy change. For now I'll say that any additional compression algorithms (bzip2/lzma/etc) will be added with a streaming interface, but since zlib.pas provides a viable alternative, adding support for it to the deflate engine isn't high on my...
2009-11-09 00:31:07 UTC by capeterson
-
Touri,
I added ExtractToStream support for CABs. Zip, tar, and gzip already supported it, as well as CreateFromStream and AddFromStream. CAB doesn't currently support either of those because the DLL's API doesn't map to it well. CreateFromStream should work for both reading an existing archive as well as creating a new one. There are some fixes for that in Subversion that weren't in the...
2009-11-09 00:03:33 UTC by capeterson
-
Craig,
Is the ExtractToStream support provided for all archive types? If so, I'll be pulling this down very soon! Also, AddFromStream support would be fantastic, if it is provided for all (possible) archive types. We typically work with ZIP, but I think anything added to Abbrevia should work with every archive type, provided the archive format itself supports doing so.
I have long asked...
2009-11-08 22:58:09 UTC by touri
-
Hello again Craig,
My apologies for a lack of clarity. By way of explanation of my intention a bit of background on this system. This is a distributed application. The initial processing of files is done by the workstation, then the prototype archives are sent to a server for integration system wide. These could be ZIp archives. At this point the server may eliminate further copies. What...
2009-11-08 21:26:37 UTC by nicoleboivin
-
Nicole,
Could you clarify what you mean by "rebuild the last cab of a spanned set"? As I mentioned previously, the CAB API does not support modifying existing CABs. AFAIK if you create a set of spanned CAB files the entire set is the archive; you can't just recreate the last file and have it work.
Craig.
2009-11-07 22:11:21 UTC by capeterson
-
Craig,
Thank you so much for your efforts so far. I will go get your changes. By way of explanation I am prototyping a rather different document management system. This is not work related, at least so far. I started this to solve my own problems with info management. I am one of those info pack-rats that has something like twenty years of projects, reference material etc and often can't...
2009-11-07 20:37:02 UTC by nicoleboivin
-
I just checked in ExtractToStream support.
AddFromStream is possible, but the compression API doesn't support it as well, so it would be a messier change. I can work on it if you need it, but probably won't bother if you don't. Loading the CAB itself from a stream or writing a new one would also be possible, but they would both be messy changes too. All three cases expect fully-qualified...
2009-11-07 05:19:25 UTC by capeterson
-
capeterson committed revision 227 to the TurboPower Abbrevia SVN repository, changing 1 files.
2009-11-07 04:55:50 UTC by capeterson
