I would like to create tags in the Visual Studio editor to insert all sorts of glyphs, adornments, text hightlightings, etc., based on line/column/length locations in the code.
I have been carefully reading the documentation walkthrough pages (https://learn.microsoft.com/en-us/visualstudio/extensibility/walkthrough-creating-a-margin-glyph?view=vs-2017 and related pages). Although a bit complex and hard to understand, it seems like the API is very much oriented on giving the means to analyse the code: it is able to give your code split into spans, with classifications, etc.
However, I have the "opposite" need: I already have the analysis done by my external analysis engine. And I already have a set of results to be displayed in the editor with line/column/length for each one. Like:
function "foo", located at line 345, column 1, length 3, and other fields containing information to be displayed,
variable "my_var", located at line 349, column 13, length 6, and other fields containing information to be displayed,
Is it possible to create tags in the Visual Studio editor directly based on their line/column/length location? Any hint, any pointer to more detailed documentation or tutorial would be greatly appreciated.
Lance's link was quite helpful to understand another way to create tags different from the MS documentation example.
Indeed, I don't analyse the text contained into the spans, the analysis is already done outside. I get some list of "defects" locations.
I get them into a defectsLocation dictionary (defectsLocation[filename][line] = location data (...)
Here is was I did:
internal class MyDefectTagger : ITagger<MyDefectTag>
{
private IClassifier m_classifier;
private ITextBuffer m_buffer;
internal MyDefectTagger(IClassifier classifier, ITextBuffer buffer)
{
m_classifier = classifier;
m_buffer = buffer;
}
IEnumerable<ITagSpan<MyDefectTag>>
ITagger<MyDefectTag>.GetTags(NormalizedSnapshotSpanCollection spans)
{
if (MyModel.Instance == null || MyModel.Instance.defectsLocation == null)
{
yield return null;
}
var filename = GetFileName(m_buffer);
if (!MyModel.Instance.defectsLocation.ContainsKey(filename))
{
yield return null;
}
foreach (SnapshotSpan span in spans)
{
ITextSnapshot textSnapshot = span.Snapshot;
foreach (ITextSnapshotLine textSnapshotLine in textSnapshot.Lines)
{
var line = textSnapshotLine.LineNumber + 1; // Lines start at 1 in VS Editor
if (MyModel.Instance.defectsLocation[filename].ContainsKey(line) &&
!MyModel.Instance.defectsLocation[filename][line].rendered)
{
var rendered = MyModel.Instance.defectsLocation[filename][line].rendered;
yield return new TagSpan<MyDefectTag>(
new SnapshotSpan(textSnapshotLine.Start, 0),
new MyDefectTag()
);
}
}
}
}
}
I need all Fields for a TFS Project Collection without specifiying a workitem. I'm already using the TFS API for a few other things, but I haven't found anything on this.
What I found was using the witadmin.exe listfields command, which gives me exactly what I want, but how do I get the results as a collection in my code? I'm not very experienced with commands, I just need the Field IDs and Names in my code so that I can display them in a WPF Listview. Do I need to call cmd.exe in my code or does the TFS API have an extension method for this? My "teacher" said it should have one since it can be done with the witadmin.exe.
witadmin.exe is a .NET Executable, so you can use Reflector, dotPeek or ilSpy on it to see how Microsoft has implemented it. This seems to be the snippet you're after:
protected void InitFields()
{
if (this.fields == null)
{
FieldDefinitionCollection definitions = new FieldDefinitionCollection(this.Store, false);
List<FieldDefinition> list = new List<FieldDefinition>(definitions.Count);
Dictionary<string, FieldDefinition> dictionary = new Dictionary<string, FieldDefinition>(definitions.Count, this.Store.ServerStringComparer);
for (int i = 0; i < definitions.Count; i++)
{
FieldDefinition item = definitions[i];
if (!item.IsInternal)
{
list.Add(item);
dictionary[item.ReferenceName] = item;
dictionary[item.Name] = item;
}
}
list.Sort(new FieldComparer(this.Store.ServerStringComparer));
this.fields = list;
this.fieldsMap = dictionary;
}
}
It will generate a list of all available fields in the Collection.
this.Store is an instance of WorkItemStore.
I'm looking for a way to automatically produce a changelog (actually a list of workitems) between two releases of my application. I have two versions of my application, v1 and v2, each is identified by a label in TFS 2010 (LABEL1 and LABEL2) that I manually created before building the setups of my app.
I have a branching system, which means I have a trunk were most of bugs are fixed, and a branch where patches are applied mostly using merges from the trunk (but there are also some fixes on the branch only that do not concern the trunk). The two versions of my application (v1 and v2) are versions from the branch.
I would like TFS 2010 to be able to return the list of bugs that were fixed (ie. the list of work items with type = Bug that are closed and verified) between these two labels.
I tried to achieve this using the web UI of TFS 2010, or using Visual Studio, but I didn't find any way.
Then I tried to ask tf.exe for a history using the following command line:
tf history /server:http://server_url/collection_name "$/project_path" /version:LLABEL1~LLABEL2 /recursive /noprompt /format:brief
where LABEL1 is the label that has been associated with the source code of the v1 of the application, and LABEL2 the label that has been associated with the source code of the v2 of the application.
It actually fails in two ways:
- the command line only returns a list of changesets, not a list of associated closed work items
- the list of changesets only contains the changesets that I applied on the branch itself, not the changesets that I also applied and the trunk and then merged to the branch. Setting or not the "/slotmode" parameter doesn't change anything.
There I tried to write a piece of C# code to retrieve the list of workitems (not the list of changesets):
var tfs = TfsTeamProjectCollectionFactory.GetTeamProjectCollection(new Uri("http://server_url/collection_name"));
VersionControlServer controlServer = tfs.GetService<VersionControlServer>();
VersionControlServer vcs = tfs.GetService<VersionControlServer>();
VersionSpec sFrom = VersionSpec.ParseSingleSpec("LLABEL1", null);
VersionSpec sTo = VersionSpec.ParseSingleSpec("LLABEL2", null);
var changesets = vcs.QueryHistory(
"$/project_path",
sTo,
0,
RecursionType.Full,
null,
sFrom,
sTo,
int.MaxValue,
true,
false); // Slotmode to false
Dictionary<int, WorkItem> dico = new Dictionary<int, WorkItem>();
foreach (Changeset set in changesets)
{
foreach (WorkItem zz in set.WorkItems)
{
if (!dico.ContainsKey(zz.Id))
{
dico.Add(zz.Id, zz);
}
}
}
foreach (KeyValuePair<int, WorkItem> pair in dico.OrderBy(z => z.Key))
{
Console.WriteLine(string.Format("ID: {0}, Title: {1}", pair.Key, pair.Value.Title));
}
This actually works, I get the list of workitems between my two labels which is actually what I wanted. But only workitems associated to changesets that were committed on the branch itself are taken into account: the workitems of type "Bug" that were solved on the trunk then merged to the branch don't appear. Slotmode doesn't change anything.
Then I finally tried to replace VersionSpecs that were defined by a label with VersionSpecs that are defined by changesets:
VersionSpec sFrom = VersionSpec.ParseSingleSpec("C5083", null);
VersionSpec sTo = VersionSpec.ParseSingleSpec("C5276", null);
And my code finally works.
So my question is: how could I get the same result with labels, which are the TFS objects I use to identify a version? If it's not possible, how should I identify a version in TFS 2010?
Thx.
Btw I found some questions on stackoverflow, but none of them gave me answers with labels. For instance:
Question example
I think http://tfschangelog.codeplex.com/ can possibly help you here.
TFS ChangeLog applicatoin allows users to automatically generate release notes from TFS. Users will have to provide information on thier project, branch and changeset range and then TFS ChangeLog application will extract information from each changeset in a given range and all the associated workitems to such changesets. i.e. it will travel from starting changeset upto ending changeset and will extract data about each changeset along with associated workitems in an XML file.
Users can then use their own transformation logic including filter, sorting, styling, output formatting, etc. to generate Release Notes Report.
Another thing I would like to add here will be related to Labels in TFS. Labels are basically assigned / associated with changesets. Currently, TFS ChangeLog application does not support Labels to define starting and ending point but it does support changeset which can be used as a workaround solution.
Hope this is useful.
In general, the absolute method of defining points in time in any SCM is clearly the checkin-id. Using labels to abstract this, is in TFS not the optimum as discussed here & here. A better approach is to use builds instead, especially in a modern CI environment.
In order to retrieve the max changeset that was contained in a given build you 'd have to do something like this:
using System;
using System.Collections.Generic;
using Microsoft.TeamFoundation.Build.Client;
using Microsoft.TeamFoundation.Client;
namespace GetChangesetsFromBuild
{
class Program
{
static void Main()
{
TfsTeamProjectCollection tpc = TfsTeamProjectCollectionFactory.GetTeamProjectCollection(new Uri("http://TFSServer:8080/Name"));
IBuildServer bs = (IBuildServer)tpc.GetService(typeof(IBuildServer));
IBuildDetail build = bs.GetAllBuildDetails(new Uri("vstfs:///..."));
List<IChangesetSummary> associatedChangesets = InformationNodeConverters.GetAssociatedChangesets(build);
int idMax = associatedChangesets[0].ChangesetId;
}
}
}
A difficulty with the above is to retrieve the BuildUri of the builds you are interested in. In order to get this information you could do something like this:
IBuildDetail[] builds = bs.QueryBuilds("TeamPorjectName", "yourBuildDefinitionName")
and then retrieve the Uri's that are important to you.
This is also a good vehicle if you eventually insist on using labels: Besides Uri, each build[] has also a LabelName.
I have been in the same situation as you. I also want Work Items from merged changesets included. I only include Work Items that are Done. Also if the same Work Item is linked to multiple changesets, only the last changeset is reported. I use this in a CI setup; and create a changelog for each build. The List<ChangeInfo> can then be exported to a XML/HTML/TXT-file. Here is my solution:
namespace TFSChangelog
{
public class TFSChangelogGenerator
{
private const string workItemDoneText = "Done";
/// <summary>
/// This class describes a change by:
/// Changeset details
/// and
/// WorkItem details
/// </summary>
public class ChangeInfo
{
#region Changeset details
public DateTime ChangesetCreationDate { get; set; }
public int ChangesetId { get; set; }
#endregion
#region WorkItem details
public string WorkItemTitle { get; set; }
public int WorkItemId { get; set; }
#endregion
}
public static List<ChangeInfo> GetChangeinfo(string tfsServer, string serverPath, string from, string to)
{
// Connect to server
var tfs = TfsTeamProjectCollectionFactory.GetTeamProjectCollection(new Uri(tfsServer));
tfs.Connect(ConnectOptions.None);
var vcs = tfs.GetService<VersionControlServer>();
// Create versionspec's
VersionSpec versionFrom = null;
if (!string.IsNullOrEmpty(from))
versionFrom = VersionSpec.ParseSingleSpec(from, null);
VersionSpec versionTo = VersionSpec.Latest;
if (!string.IsNullOrEmpty(to))
versionTo = VersionSpec.ParseSingleSpec(to, null);
// Internally used dictionary
var changes = new Dictionary<int, ChangeInfo>();
// Find Changesets that are checked into the branch
var directChangesets = vcs.QueryHistory(
serverPath,
VersionSpec.Latest,
0,
RecursionType.Full,
null,
versionFrom,
versionTo,
Int32.MaxValue,
true,
false
).Cast<Changeset>();
foreach (var changeset in directChangesets)
{
foreach (var workItem in changeset.WorkItems.Where(workItem => workItem.State == workItemDoneText))
{
if (changes.ContainsKey(workItem.Id))
{
if (changeset.ChangesetId < changes[workItem.Id].ChangesetId) continue;
}
changes[workItem.Id] = new ChangeInfo { ChangesetId = changeset.ChangesetId, ChangesetCreationDate = changeset.CreationDate, WorkItemId = workItem.Id, WorkItemTitle = workItem.Title };
}
}
// Find Changesets that are merged into the branch
var items = vcs.GetItems(serverPath, RecursionType.Full);
foreach (var item in items.Items)
{
var changesetMergeDetails = vcs.QueryMergesWithDetails(
null,
null,
0,
item.ServerItem,
VersionSpec.Latest,
0,
versionFrom,
versionTo,
RecursionType.Full
);
foreach (var merge in changesetMergeDetails.Changesets)
{
foreach (var workItem in merge.WorkItems.Where(workItem => workItem.State == workItemDoneText))
{
if (changes.ContainsKey(workItem.Id))
{
if (merge.ChangesetId < changes[workItem.Id].ChangesetId) continue;
}
changes[workItem.Id] = new ChangeInfo { ChangesetId = merge.ChangesetId, ChangesetCreationDate = merge.CreationDate, WorkItemId = workItem.Id, WorkItemTitle = workItem.Title };
}
}
}
// Return a list sorted by ChangesetId
return (from entry in changes orderby entry.Value.ChangesetId descending select entry.Value).ToList();
}
}
}
This question got me closer to solving a similar problem I was having.
Use the type LabelVersionSpec instead of VersionSpec for label versions.
Replace:
VersionSpec sFrom = VersionSpec.ParseSingleSpec("LLABEL1", null);
VersionSpec sTo = VersionSpec.ParseSingleSpec("LLABEL2", null);
with:
LabelVersionSpec sFrom = new LabelVersionSpec("LLABEL1");
LabelVersionSpec sTo = new LabelVersionSpec("LLABEL2");
I have been working on implementing a custom RSS feed in sitecore 6.4. My custom behaviour is very limited, all i effectively wanted to is add a link for author (our author field is a reference field so we cannot use the built in author attribution).
I overrode RenderItem() on the PublicFeed class so that i could make use of my own implementation of the FeedRenderer class (where the author logic is housed). my approach follows this pattern outlined by John West for adding your own rendering behaviour:
public class MyPUblicFeed: PublicFeed
{
protected override SyndicationItem RenderItem(Item item)
{
Assert.ArgumentNotNull(item, "item");
Control rendererControl = FeedUtil.GetFeedRendering(item);
if (rendererControl == null)
{
return null;
}
using (new ContextItemSwitcher(item))
{
var myRenderer= rendererControl as MyFeedRenderer;
if (myRenderer!= null)
{
myRenderer.Database = SitecoreHelper.CurrentDatabase.Name;
return myRenderer.RenderItem();
}
var renderer = rendererControl as Sitecore.Web.UI.WebControls.FeedRenderer;
if (renderer != null)
{
renderer.Database = SitecoreHelper.CurrentDatabase.Name;
return renderer.RenderItem();
}
}
throw new InvalidOperationException("FeedRenderer rendering must be of Sitecore.Web.UI.WebControls.FeedRenderer type");
}
}
And now for my rendering class:
public class MyFeedRenderer: Sitecore.Web.UI.WebControls.FeedRenderer
{
public override SyndicationItem RenderItem()
{
Item item = base.GetItem();
var syndicationItem = base.RenderItem();
//unfortunately we have to parse params again :(
FeedRenderingParameters feedRenderingParameter = FeedRenderingParameters.Parse(base.Parameters);
AddAuthor(syndicationItem, item, feedRenderingParameter);
return syndicationItem;
}
private static void AddAuthor(SyndicationItem syndicationItem, Item item, FeedRenderingParameters feedRenderingParameter)
{
//clear out authors added by base class
syndicationItem.Authors.Clear();
//logic for adding author here
}
}
this all works great, outputting exactly what i want, but the caching element doesn't appear to be working. I have set the cacheable flag on the actual item itself with a timespan of 01:00:00. This didn't appear to work - if i put a breakpoint in either of the above classes it is hit everytime the feed is requested.
so then i tried to enable caching at a control level, turning caching on with VaryByData for the MyFeedRenderer rendering. alas this isn't working either, the breakpoint is hit every time.
Can anyone offer any advice on this matter? the documentation simply recommends turning it on on the actual feed item, not at the Rendering level, but neither seem to be working for me. Interestingly HTML caching is working elsewhere - is RSS also put into the HTML cache?
Thanks in advance,
Nick
-Ensure the Cacheable checkbox in the feed definition item is checked.
-Ensure that you have published the feed definition item.
-If you do not populate the Cache Duration field in the feed definition item, it should default to one day.
-Feeds appear to cache in Sitecore.Syndication.FeedManager.Cache rather than the site output cache. Inspect that cache object in the Visual Studio debugger after calling your feed, and then again after calling that feed a second time, to try to see if any records appear, and if multiple cache keys appear for the same feed. Investigate the Render() method; if PublicFeed.IsCacheable() returns false (depending on the Cacheable field in the feed definition item), PublicFeed.Render() does not cache.
-Ensure nothing else clears caches between your requests for the feed.
SDN forum thread: http://sdn.sitecore.net/forum/ShowPost.aspx?PostID=40591
I'm hoping that I'm just using the API incorrectly, but for some reason, when I query my oData service from wp7, I am only receiving the last entry in the sequence. I'm pretty sure the service is working just fine as I'm able to just visit the oData query URL in the browser and get the correct results. However, the following method only ends up displaying one item on the list.
I based the following code from the sample at this blog post: http://chriskoenig.net/2010/10/30/odata-v2-and-windows-phone-7/
The observable collection which is passed in is the one which is bound to the wpf listbox.
private static void LoadRuntimeData(ObservableCollection<CategoryItem> items)
{
var catalog = GetCatalog();
var queryUri = new Uri("/Categories?&$orderby=Name", UriKind.Relative);
var categories = new DataServiceCollection<Category>(catalog);
var queryHandle = App.ViewModel.StartQuerying();
categories.LoadAsync(queryUri);
categories.LoadCompleted += (e, c) =>
{
using (queryHandle)
{
var serverItems = categories.Select(k => new CategoryItem
{
Name = k.Name,
Description = k.Description
});
items.Clear();
foreach (var item in serverItems)
{
items.Add(item);
}
}
};
}
Any tips would be greatly appreciated.
Edit: per a comment below, I've uploaded the source code in which this issue is reproducible: http://khanviewer.codeplex.com/
Taking a quick stab at this (and I suspect that Chris would know better).
But don't you need to somehow identify/tag with an attribute a primary key column (the property or properties that define uniqueness).
I know I have run into something similar to this with RIA Services, and the issue there was that I wasn't setting a value for the unique identifier column. As a result I saw all my data come down, but the when the client got the data it saw all the data as a single record.
Jay