Locations starting with another one's name not navigable

[vwbij]

Hi,

The selection of the current repository entry in querySnapshot was broken. It finds a match for "KEY" in "KEY_AB", "KEY_DE" and so on. Here the fixed loop head:

while (loc)
{
    const int minLen = min(loc->title.len, pathLen);
	if (!strncmp(path, loc->title.data, minLen))
    {
        apr_pool_t *subPool = NULL;
        svn_error_t *err;
        svn_opt_revision_t revision;
        size_t subPathLen = 0;
        char *buf = NULL;
        strbuf_t s;

		// Lets look if this is correct. If the path is longer than the title, we need to see
		// a path seperator next (\ or /). If not, we have it to do with a matching part of a name
		// ("UDO_NEW" instead of "UDO"). Good match would be: "UDO/DevGuide" or "UDO\DevGuide".
		if(pathLen>minLen)
		{
			if(!((path[minLen]=='/')||(path[minLen]=='\\')))
			{
		        loc = loc->next;
				continue;
			}
		}

        /* found it, fetch data from SVN */
        subPool = svn_pool_create(Subversion.pool);
        subPathLen = strlen(path + minLen);
        buf = (char*)apr_palloc(subPool, loc->url.len + subPathLen + 1);
		s.data = buf;
		s.size = loc->url.len + subPathLen + 1;

All the Best,

Volker

[mvf]

Hi Volker,

sorry for the late reply and thanks for taking the time to debug and report this. Unfortunately that part of the code seems broken on a conceptual level, so the proposed change is unlikely to fully fix it. At the very least there are other parts that rely on the same faulty reasoning. The root cause is understood and I'll look into it the next time I boot Windows.

Cheers,
Matthias

[vwbij]

Hi Matthias, Thank you. Yes, you are right, this was a conceptual problem, I've found the same problem on other corners. I have transfered the whole code to C++ - I am poor in plain C programming - and have fixed some problems: * Broken special characters (filenames!): SVN uses always UTF8, need to transfer between the current ANSI codepage in the UI and UTF8 in the SVN core * FsContentGetValue: the snapshot was re-build very often, I could not see the reason why. Deactivating the condition "!snapshot->subPath.data" leads to a situation where all files of a folder was enumerated from the same snapshot - as far as I can see this was the way it was planed(?). I found a problem with the file size. Some SVN servers seem to report wrong file sizes. This is a problem because it kills the ability to use the TC synchronize function to detect changed files. It would work to compare with a local copy, if you set the checkbox "ignore timestamps". But it depends on the possibility to have the proper size. So, maybe you can help me to answer the following questions: * Is it possible to retrieve the "real" filesize from SVN without taking a temporary copy of the file (very slow)? * Is it possible that this behaviour has changed with the version of the SVN server? * Is it possible to make "compare by file content" work with the plugin? It is not working - reports just all files changed. Now, enough questions. No problem if you are not able to answer them - have to research more. I give you my source as I've finnished it. Maybe we can make a brunch from it? Cheers, Volker Am 01.02.2017 um 18:45 schrieb Matthias von Faber:

…

Hi Volker, sorry for the late reply and thanks for taking the time to debug and report this. Unfortunately that part of the code seems broken on a conceptual level, so the proposed change is unlikely to fully fix it. At the very least there are other parts that rely on the same faulty reasoning. The root cause is understood and I'll look into it the next time I boot Windows. Cheers, Matthias — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AVVXUXPE570Txwzc_4QL6_tlKnLil6Xwks5rYMTLgaJpZM4LllX1>.

[mvf]

Hi Volker,

thanks for reporting all those points. I have taken the liberty of creating a separate issue for each one and I kindly ask that you do the same for any new topics that come up, regardless of whether it's a defect, request, suggestion, etc. This will make it much easier to keep track of things.

In general, patches are welcome of course, but I don't see any benefits in using C++. C99 is fine now though, since it's finally supported by Microsoft compilers.

Thanks,
Matthias

[vwbij]

Hi Matthias, Thank you for dividing the issues into seperate threads, I know it is not good to have multiple issues in one message. I know, C99 is fine, but it is not handy for me. So I've decided to convert the project to reach better comfort and faster progress while programming. I'd like to share my work with anybody who wants, as soon as I have a stable working state. Sorry, you may have the same issue with my C++ code like I have with plain C99 :-(... Thanks, Volker Am 05.02.2017 um 13:37 schrieb Matthias von Faber:

…

Hi Volker, thanks for reporting all those points. I have taken the liberty of creating a separate issue for each one and I kindly ask that you do the same for any new topics that come up, regardless of whether it's a defect, request, suggestion, etc. This will make it much easier to keep track of things. In general, patches are welcome of course, but I don't see any benefits in using C++. C99 is fine now though, since it's finally supported by Microsoft compilers. Thanks, Matthias — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AVVXUVKRIoa8nq9OhC7yZIqUN0ad4lEnks5rZcJ2gaJpZM4LllX1>.

[mvf]

Hi Volker,

sorry if my C++ comment ended up sounding a bit hostile, I'm not opposed to C++ in general. I just don't see how it would help me personally in this somewhat low-level project. If you are more productive with C++, more power to you! 😉

One thing that's perhaps worth mentioning: ATM this program is even C89 because that's all Visual Studio supported at the time. Even now I remember that as extremely irritating. Not being able to mix declarations and code was the biggest issue because it degrades function structure. So on the off-chance you're using a version of Visual Studio older than Visual Studio 2013, upgrading would enable a useful subset of C99.

Happy hacking,
Matthias

[vwbij]

Hi Matthias,

FYI / offtopic: I have finished a first version and send it to some fellows for test. You may download it from "https://dl.dropboxusercontent.com/u/15963916/TCP.ZIP" if you want. Source is not included yet, have to cleanup the code a little bit :-). Main improvement is the ability to get the "real" file size.

All the Best, Volker